非傳統學術指標的比較研究：以國立臺灣大學文學院的研究產出為例

Abstract

鑒於傳統引文分析法無法有效評估人文學與社會科學領域的學術影響力，加上以網際網路進行學術傳播的方式蓬勃發展，學術影響力的意義不再侷限於期刊文章被引，新興學術評估指標Altmetrics以網際網路社群媒體分析學術影響力的方式，則能夠補足引文分析的不足，本研究以非傳統學術指標全面分析人文學與社會科學學者在Altmetrics研究產出的特性及差異。 研究對象為臺灣大學文學院243位學者，包含37位教授、14位副教授、6位助理教授，透過網際網路蒐集報章雜誌被引數、線上課程大綱被引數、維基百科被引數、Google學術搜尋被引數、Google學術搜尋著作數、社群媒體被引數、指導論文數、博碩士論文中被引數、Mendeley著作數、Mendeley被關注數 10項學術表現資料來源，及學者獲獎數及計畫數2項學術成就資料來源，蒐集時間從2014年10月至2015年11月。以無母數統計檢定、集群分析、相關分析、零膨脹負二項迴歸模型等統計分析方法，探討變項間之關係。 研究結果顯示：（1）多數資料來源彼此間具有顯著相關，能過作為Altmetrics指標的潛力，尤其是Google學術搜尋資料來源。（2）人文學科、社會科學學科與藝術科學學科在不同Altmetrics資料來源的表現上有顯著差異，人文學者較容易出現在雜誌上，藝術學者於報章雜誌及社群媒體資料來源學術表現上佳，社會科學學者於學術社群媒體平台活躍程度較人文及藝術學者高。（3）通常研究與教學影響力指標較高的學者，社會大眾能見度的指標也會獲得較高的影響力（4）本研究所蒐集資料來源對學術成就解釋力不佳，但指導論文數、博碩士論文中被引數、報章雜誌被引數能夠解釋學術成就。 目前進行學術評估時，時常將人文學、社會科學、藝術領域納入一起評估，但透過本研究結果，三者之間具有差異，因此進行學術評鑑時，不應以相同標準進行評估。

Given that traditional citation analysis cannot monitor the research productivity of scholars in the social sciences and humanities effectively, and with the development of the internet, the use of social media and the internet in research communication has become more and more popular. As a result, the meaning of a scholar’s academic influence is not confined to the citation counts for the papers in journals anymore. Altmetrics, a new proposed method for evaluating academic performance, can access not only the research impact but also social impact of scholars, and can resolve the limitation of citation analysis. The purpose of this study is using altmetrics to analyze and compare the differences and characteristics of research outputs in the social sciences and humanities. A total of 243 researchers in the College of Liberal Arts, National Taiwan University, were studied, including 37 professors, 14 associate professors and 6 assistant professors. Ten data sources of academic performance, including newspaper and magazine citations, syllabus citations, Wikipedia citations, Google Scholar citations, the numbers of papers in Google Scholar, social media citations, the number of advised theses and dissertations, theses and dissertations citations, Mendeley reader counts, the numbers of papers in Mendeley, and 2 data source of academic achievement, including prize counts and project counts were collected through the internet. The time of data collection was from October 2014 to December 2015. Nonparametric statistics, cluster analysis, correlation analysis and zero-inflated negative binomial regression were used for statistical analysis. The major results are summarized as follows. First, there are weak or moderate, but the significant correlation between data sources, which suggest that these data sources have potential to be used as altmetric indicators, especially Google Scholar. Second, there are significant differences between scholars in the social sciences, humanities and fine arts in terms of performance in different altmetric data sources. Humanities scholars are more likely to appear in magazines than social science scholars. Fine arts scholars have better performance in newspapers, magazines and social media. Social science scholars are more active in academic social networking platforms. Third, the higher the performance of scholars as measured by the indicator of influence on research and teaching, the higher the performance of scholars in terms of the indicator of social visibility. Fourth, some data sources, including newspaper and magazine citations, the number of advised theses and dissertations, as well as theses and dissertations citation, has some explanatory power for the academic achievement of scholars. Finally, the results suggest that we should not put scholars in the humanities, social sciences and fine arts together in research evaluations because of their differences in citation behavior.