本研究以美國國家科學基金會所建置之「結案備忘錄資料庫」為研究工具，檢索1989年至2019年間共3,064起疑似研究違失案件；分析項目包含案件所涉之行為類型與採取的後續行動，以及兩者在過去31年間之消長。主要結果包括：歷年之疑似研究違失案件共分31種行為類型與5種後續行動；結案總案件數在2004年後顯著增加，並於同年達到高峰；結案案件中以逐字抄襲為最大宗；在後續行動上，情節較輕的案件多是向當事人進行警告，情節較嚴重的案件則最常被施予行政處分，其次則是追回經費與被判定為不當研究行為。本文依研究結果，針對中央主管機關在適度公開案件資訊及建置案件資料庫等面向，提出相關建議與作法。

The present study concerned the database, Case Closeout Memoranda, established by the National Science Foundation (NSF) of the United States. In the study, the current authors retrieved 3,064 reported cases of suspected noncompliance in research between 1989 to 2019. We then analyzed the classification of misbehaviors and subsequent actions; the authors also investigated the evolution of the violations and actions over the three decades. The preliminary results include the following: the closed cases over the years can be categorized into 31 types of misbehaviors and 5 types of subsequent actions; the total number of closed cases increased significantly after 2004 and reached a peak in the same year; among the closed cases, verbatim plagiarism accounts for the most significant reported breach, followed by grant frauds; in terms of subsequent actions, individuals with fewer breaches were often warned about questionable research practices, and a major portion of those with more severe issues performed administrative actions, followed by monetary recovery and being defined as research misconduct. Finally, the authors proposed establishing a national case database of noncompliance in research in Taiwan to promote retrospective studies and categorization research further. In addition, the authors hoped that Taiwanese local government ministries and research institutions can consider the suggestions in the article to uphold accountability, transparency, and social responsibility.