本研究以Scientometrics期刊論文的參考文獻資料，對科學計量學領域的特性進行分析。過去的研究裡，Schubert and Maczelka (1993)利用1980-1981和1990-1991兩個時段Scientometrics期刊論文的參考文獻年齡，計算兩個時段的Price指標(Price index)，發現後面時段的Price指標較大於前一個時段的Price指標，因此判斷科學計量學領域正趨向較硬的社會科學。Wouters and Leydesdorff (1994)認為只取樣幾個時段的資料可能不準確，應該觀察資料的連續變化情形，他們的研究指出Price指標在1978到1992年間的變化並不大，所以這個社會科學是較穩定的、並沒有朝向應科學發展的趨勢。Glänzel and Schoepflin (1994)則認為科學計量學是異質性很高的研究領域，這個領域的作者來自不同的學科，擁有相當不同的傳播、引用和發表方式，這種現象使得這個科際整合的領域趨向多元發展。
本研究以1980、1989和1997三個年度的Scientometrics期刊論文的參考文獻做為研究資料，計算以下四個統計資訊：1) 每篇論文的Price指標(the Price index per paper)、2) 參考文獻是連續出版品的百分比(the percentage of references to serials)、3) 參考文獻的平均年齡(the mean reference age)和4) 平均參考文獻數(the mean reference rate)。各年度的論文數與指標如下表所示：
三個年度的期刊Price指標相當穩定，並且所有的指標都反映Scientometriics類似社會科學期刊的模式。此外，表上也呈現每一年度論文的Price指標和連續出版品比率的中位數(median)，這些數據皆明顯大於期刊的數據，表示許多論文比整體期刊呈現的情形來得硬。
本研究並分派每一篇論文到一個類別，這些類別包括：1) 書目計量學理論、數學模型與書目計量學定律的公式化(bibliometric theory, mathematical models and formalisation of bibliometric laws)；2) 案例研究與實務論文(case studies and empirical papers)、3) 方法學論文包含應用(methodological papers including applications)、4) 指標工程與資料呈現(indicators engineering and data presentation)、5) 書目計量學中的社會學取向，科學社會學(sociological approach to bibliometrics, sociology of science)、6) 科學政策、科學管理與廣泛或技術討論(science policy, science management and general or technical discussions)。論文在各類別及各年度的分布如下表所示
可以明顯地發現：1980年度的案例分析和方法學論文所占比例較小，但在後兩個年度則占了論文的大部分；相反的，科學政策從第一個年度後便減少，社會學取向也有相同的情形。本研究推測等Research Evaluation與Social Studies of Science等期刊吸引了相關的論文投稿可能是科學政策與社會學取向方面的論文數量減少的原因。
下表分析各分類論文的參考文獻指標，從這些數據可以看出不同類別的論文在指標上差異很大，特別是科學政策，因此當科學政策在第二和三個年度大幅減少後，期刊整體的指標便有很大的不同。因此，本研究認為科學計量學是異質性很高的領域。
The development of the field of bibliometric and scientometric research is analysed by quantitative methods to answer the following questions:
(1) Is bibliometrics evolving from a soft science field towards rather hard (social) sciences (Schubert-Maczelka hypothesis)?
(2) Can bibliometrics be characterised as a social science field with stable characteristics (Wouters-Leydesdorff hypothesis)?
(3) Is bibliometrics a heterogeneous field, the sub-disciplines of which have their own characteristics? Are these sub-disciplines more and more consolidating, and are predominant sub-disciplines impressing their own characteristics upon the whole field (Glänzel-Schoepflin hypothesis)?
The findings suggest, that the field is in fact heterogeneous, and each sub-discipline has its own characteristics.
Indeed, this journal covers almost the complete spectrum of bibliometric research. It publishes theoretical papers and papers on mathematical models as well as on the research evaluation of special fields and/or selected institutions, on science policy questions as well as articles on social studies of science and general discussions about the field.
While Schubert and Maczelka (1993) found a clear move from softer’ towards ‘harder’ (social) sciences between the analysed time periods 1980-1981 and 1990-1991, respectively, Wouters and Leydesdorff (1994) concluded on the basis of the change of Price’s Index in time that bibliometrics has not become a hard social science field in the observation period 1978-1992.
Glänzel and Schoepflin (1994) stated in their discussion paper that bibliometrics has become a heterogeneous field and sub-disciplines are drifting apart. Consequently, bibliometrics comprises sub-disciplines with distinctly different communication, citation and publication characteristics.
Assuming that bibliometrics is an interdisciplinary field and that authors coming from different fields bring their specific communication behaviour into it, we have classified all papers published in Scientometrics into different categories representing the main field-specific approaches to bibliometrics.
All source articles published in the journal Scientometrics in three sample years, 1980, 1989 and 1997, have been processed. All references cited in articles, notes and letters in the above three publication years were selected. Review articles have not been taken into consideration since the extent and structure of the reference lists of these documents are expected to distinctly differ from those of other research papers. Papers without references have been omitted.
References have been assigned to two categories, reference to serials (S) and reference to non-serials (N). All references have been classified manually.
The following statistics have been calculated.
1. The Price Index per paper. This index is defined as the percentage of references not older than five years in all references of an individual paper. This indicator has been introduced by Moed (1989).
2. The percentage of references to serials. The share of references assigned to category S in all references (N+S) cited by a journal or subfield expressed in percent.
3. The mean references age. The age of references cited in a journal or subfield are summed up and divided by the number of the references. This indicator can be determined also as a conditional mean, that is for both the subset of references in serials and non-serials separately.
4. The mean reference rate. This is the ratio of the number of references cited by a journal and the total number of papers published in the journal including those without references.
The “Price Index” is commonly used as a measure to distinguish between hard science, soft science, technology and non-science (see Price, 1970).
According to the results of an earlier study (Glänzel and Schoepflin, 1999), the percentage of references to serials proved to be a sensitive measure to characterise typical differences in the communication behaviour between the sciences and the social sciences.
The mean reference age also serves as an efficient measure of the “hardness” of science. In the paper by Glänzel and Schoepflin (1999), a comparison of the mean age of references and the Price Index has shown that the age of references is only in part reflected by the Price Index, in particular if the average age of references does not exceed about 15 years.
In addition, we calculated the mean reference rate, that is, the average size” of the reference list of a bibliometric paper published in Scientometrics. Although this cannot be considered a sensitive measure of the “hardness” of science, it reflects nevertheless additional field-specific characteristics (see Glänzel and Schoepflin, 1999).
In the next step, all selected source articles in Scientometrics have been assigned manually to one category out of a scheme of six. ... The classification scheme used for this study is presented in Table 1. The classification permits to group the material in several ways: the categories can be regarded from the viewpoint of core bibliometrics (2, 3, and 4) and background research (1, 5, and 6), but also with respect to theoretical (1, 3, and 5) and applied research (2, 4, and 6).
Table 2 presents the number of papers assigned to each category in 1980, 1989 and 1997.
However, the Price Index shows stability also in our case. Even more, all indicator values reflect patterns typical of social-science journals (c.f. Glänzel and Schoepflin, 1999).
It is worth mentioning here, that the median of paper-based indicators is greater both for the Price Index and the share of serials than the corresponding journal indicators. This phenomenon allows only one possible interpretation: Numerous papers are ‘harder’ than expected on the basis of the overall journal indicators.
There are obviously two dramatic developments: first, there is an impressing and steady growth of Case Studies, from a forth position in 1980 to the predominant first position in 1997. Second, there is a similarly impressive loss of share of articles on Science Policy and Discussions (category 6) from the predominant first position in 1980 to a minor category in 1997. This goes along with a steady loss in material with a sociological approach, too (category 5). On the other hand, there is a certain increase in Methodology (category 3), while Theory and Indicator Engineering remain minor classes.
If we now take a look on Core Bibliometrics as defined by categories 2, 3, and 4, it becomes obvious, that this group is practically reduced to Case Studies and Methodology, and by far dominating the total output of research as represented by the journal Scientometrtics.
The group characterised as Background Research (categories 1, 5, and 6) has continuously lost ground since 1980. Moreover, theoretical research in bibliometrics seems to be mainly a matter of methodology.
Following the differentiation of the field, journals like Social Studies of Science or the newly founded Research Evaluation publish a considerable share of bibliometric research in categories 5 and 6. But also journals in information science (e.g. JASIS, Information Processing & Management, or Journal of Information Science to name just a few Anglo-Saxon titles) attract bibliometric research articles.
On the detriment of a larger scope, Scientometrics has clearly become the forum for Case Studies and Methodology-oriented contributions.
The above-mentioned deviating patterns in 1980 can be at least in part explained by the great share of papers in category 6 and their low share of serials and relatively low age of the references. Indeed, the indicator values of the six categories give evidence of specific characteristics of the corresponding sub-disciplines.
The above trends and figures tell unambiguously against an evolution of bibliometrics towards a discipline of ‘hard’ social science (Schubert-Maczelka hypothesis). On the other hand, we cannot speak of stable characteristics, either (Wouters-Leydesdorff hypothesis). The indicators allow only an interpretation in the sense of the third hypothesis. The field is indeed heterogeneous, and each sub-discipline has its own characteristics. This may, of course, be at least in part caused by the deviating field-specific communication behaviour of the authors who bring traditional organisation schemes from their own fields into bibliometrics.
沒有留言:
張貼留言