撤稿對作者文章出版、學術影響力與合著規模之影響

Abstract

本研究採用書目計量法，第一部分蒐集Web of Science資料庫中所有2017年前出版之撤稿論文，透過分析撤稿論文的年代分布、撤稿時間差、國家分布、領域分布、平均作者數、被引次數、以及作者撤稿次數分布等，探討撤稿現象之整體特性和趨勢。第二部分探討遭撤稿作者在撤稿前後之學術表現變化。本研究依撤稿次數將作者分為單篇撤稿作者與多篇撤稿作者，採用不等比例分層隨機抽樣法，分別抽取361位單篇撤稿作者與244位多篇撤稿作者，並進一步蒐集WoS資料庫中這605位撤稿作者在撤稿前後5年出版之論文，從文章出版、學術影響力與合著規模三面向比較兩組作者在撤稿前後之學術表現變化。 第一部分研究結果顯示，撤稿論文數於近年快速成長，近十年（2008-2017）的撤稿論文數為前十年（1998-2007）的七倍。大多數撤稿論文來自臨床醫學和生命科學領域。約28.3%的撤稿論文來自美國作者，中國作者其次，占24.6%。近十年中國作者之撤稿論文數大幅成長，甚至超越美國作者之撤稿論文數。在作者的撤稿次數分布上，大多數的作者為單篇撤稿作者（88.6%），僅11.4%的作者為多篇撤稿作者，極少數作者有大量論文遭到撤稿，最高撤稿次數為68篇撤稿論文。在撤稿作者的國家分布上，約31.8%的單篇撤稿作者來自中國，在所有國家中排名第一；而約28%的多篇撤稿作者來自美國，在所有國家中排名第一，其次為中國，占多篇撤稿作者約20.4%。此外，分析多篇撤稿作者佔各國所有撤稿作者之比例發現，日本國內的多篇撤稿作者比例最高（20.8％），其次為美國（16.1%）和伊朗（15.8%），表示這些國家的撤稿作者有較高比例為兩篇以上撤稿的作者。 第二部分研究結果顯示，在文章出版方面，撤稿後多篇撤稿作者相較單篇撤稿作者有較高比例不再繼續發表學術論文；在出版文章數上，撤稿後多篇撤稿作者之論文數量顯著較低，單篇撤稿作者則沒有顯著變化。從學術影響力分析，兩組作者撤稿前後之學術影響力均無顯著變化。在合著規模上，分析結果顯示兩組作者在撤稿後文章作者數之平均數與中位數均顯著增加；而個人合著網絡規模的分析結果顯示，撤稿後單篇撤稿作者之個人合著網絡規模顯著較大，多篇撤稿作者則沒有顯著變化。本研究進一步發現多篇撤稿作者存在組內差異，兩篇撤稿的作者在撤稿後個人合著網絡規模並無顯著差異，但三篇以上撤稿的作者則在個人合著網絡規模顯著較小。該研究結果顯示，雖然撤稿作者在文章平均作者數上並無顯著變化，在個人合著網絡規模上，可看到不同組別作者撤稿前後表現存在差異，撤稿次數較高之作者在撤稿後總體合著的作者人數確實較少。

This study employs bibliometrics and the objectives of this study are twofold. First, we examine the overall characteristics and trends of retracted articles in the Web of Science database. Second, we investigate the pre- and post-retraction performance of retracted authors in terms of productivity, scientific impact, and coauthorship size. Systematic comparisons are made between single-retraction authors and multiple-retraction authors to explore the possible effects of scientific retractions. In the first part of this study, results revealed that retractions are a recent and growing phenomenon, with a dramatic growth of more than sevenfold in 2008–2017 compared to 1998-2007. Most retracted articles are from the fields of Clinical Medicine and Life Sciences. While authors from the U.S. account for the largest proportion of retracted articles, authors from China contribute to a fast-growing number of retracted articles. Single-retraction authors accounted for the vast majority of authors (88.56%), while multiple-retraction authors accounted for 11.44% of all authors in this study. China had the highest number of single-retraction authors; the U.S. ranked first in terms of the number of multiple-retraction authors. Furthermore, Japan had the highest proportion of multiple-retraction authors (20.8%) among its total number of retracted authors, followed by the U.S. (16.1%) and Iran (15.8%), suggesting that retracted authors from these countries were more likely to receive two or more retractions compared to other countries. In the second part of this study, results revealed that single-retraction authors and multiple-retraction authors showed differing trends in terms of productivity and the overall number of collaborators, but not with respect to scientific impact and the number of authors per paper. Specifically, single-retraction authors were more likely to have continued publishing after the retraction. From the pre-retraction to the post-retraction period, single-retraction authors showed no observable change in publication count or scientific impact, an increase in both the number of authors per paper and the overall number of collaborators in their personal coauthorship network. By contrast, multiple-retraction authors were more likely to discontinue publishing after retraction. From pre- to post-retraction, multiple-retraction authors had a significant decrease in publication count, no change in scientific impact, an increase in the number of authors per paper, and no change in the overall number of collaborators. Different from previous research, this study took a first look at the scientific impact of post-retraction publications by retracted authors, and found comparable citation rates between the author’s pre- and post-retraction publications for both single-retraction and multiple-retraction authors. This finding was rather unexpected, and suggests that retracted authors may experience no severe consequences in terms of scientific impact despite receiving one or even multiple retractions. One of the novel contributions of this study is a deeper look into the coauthorship size of retracted authors. We took an approach based on personal network analysis, and observed differential trends for the overall number of collaborators from pre- to post-retraction between the two groups of authors. Furthermore, this study revealed some variance within multiple-retraction authors—authors with 3 or more retractions had a significantly smaller collaboration network after retraction, while that of authors with 2 retractions remained steady from pre- to post-retraction. This suggests that while retractions seemed to have no impact on coauthorship size based on the average number of authors per paper, a look from the perspective of the author’s personal coauthorship network revealed that authors with a higher number of retractions did seem to experience more severe consequences after the event of retraction.