A retraction is a journal's statement informing the scientific community that a previously published study is not reliable, or ethical and should no longer be cited. This retraction is formal annoucement that the paper should be removed from the literature consulted by scholars during reaseach.
Once an article is retracted it usually marked in some way in most databases.
Example of an retracted Article:
According to the Committee on Publication Ethics (COPE ) Retraction Guidelines Version 2: November 2019
Editors should consider retracting a publication if:
• They have clear evidence that the findings are unreliable, either as a result of major error (eg, miscalculation or experimental error), or as a result of fabrication (eg, of data) or falsification (eg, image manipulation)
• It constitutes plagiarism
• The findings have previously been published elsewhere without proper attribution to previous sources or disclosure to the editor, permission to republish, or justification (ie, cases of redundant publication)
• It contains material or data without authorisation for use
• Copyright has been infringed or there is some other serious legal issue (eg, libel, privacy)
• It reports unethical research
• It has been published solely on the basis of a compromised or manipulated peer review process
• The author(s) failed to disclose a major competing interest (aka, conflict of interest) that, in the view of the editor, would have unduly affected interpretations of the work or recommendations by editors and peer reviewers.
Journals are notified of infraction by editors of the journal, author of the article, or readers of the article.
An investigation is initiated and if warrented follow by a the publication of a retraction notice by editor, publisher,or learned society.
The article is then marked and retracted.
Retraction Watch Database (retractiondatabase.org)
Welcome to our database. We’ve prepared this document to help you get started, and to answer some questions that are likely to come up. This document will evolve as users have more questions, so please feel free to contact us at team@retractionwatch.com.
But before you begin – a couple of important notes to keep in mind:
Preprints, guidelines, and reviews (such as those published in the Cochrane Database of Systematic Reviews) often undergo several revisions which can be overwritten on the original HTML page. Each revision may have its own DOI, which can lead to a problem for those hoping to keep track of the status of a given article. For example, the first version published may be given a DOI, and updated versions may overwrite the same html page but have a different doi – usually with something like “.v2” or ”. pub2” after the original DOI. However, if a retraction is later published, the retraction will then apply to all versions, including the original article. As we do not currently have the ability to include several DOIs in the same entry, users who store later versions in digital libraries risk their stored version not being flagged as retracted.
To address such problems, we are entering the original article as well as the revision(s) into the RWDB, provided the revisions have a DOI. This will allow people with different versions in their digital library to be aware of the retraction. However, this can also falsely inflate retraction numbers in general, and more specifically for journals and for the authors themselves. To help alleviate this issue, users should look for the term “Revision” in the article type. The term “Revision” will be added only to the revised versions of the original article. Consideration should be given as to whether to include or omit the revised versions from retraction counts based on the user’s intentions. In this way, the RWDB can be useful to those who need it to maintain the current status of their bibliography as well as to those researchers who tally counts of types of retractions for journals, authors, and other criteria.
If you want to see the total number of entries in the database, click on the green Search button. Answer “OK” to the question prompt. The maximum number of entries that the database will show is 600, but the total number of entries will show in a yellow banner above the results.
Keep in mind that this search will return not just retractions, but also the corrections and expressions of concern in the database. (See Question 11.) That means you’ll want to limit your search to retractions — using the “Nature of Notice” dropdown — if you just want a total number of retractions for a particular author, or country, or journal, etc.
Whenever you want to change your search, just click on “Clear Search”, which will clear all search fields.
In general, if you’re doing any kind of research project, we recommend you consult with us to make sure you know what is — and what isn’t — possible given what’s in the database. We’d love to help, so contact us at team@retractionwatch.com.
You can type in any or all fields. The more fields you fill in, the more specific the search; the fewer fields you use, the more general. For example, if you are looking for the number of retractions in a particular journal during a particular time frame, you can complete the Journal, Nature of Notice, and Dates fields to narrow your results.
Using multiple fields defaults to the Boolean “AND” for each field. Searches for multiple words in the same field use Boolean criteria: “AND”, “OR”, and “AND NOT”. You can also use a wildcard character (*) before or after your selected word/phrase to enlarge your search parameters.
Some fields (for example: Journal) have drop-down menus signified by a little “arrow box” at the end of the field. You can either type in your choice or select from the drop-down menu. Fields without a drop-down menu require a free text entry – just type keywords or phrases. A serious caution here: Skip the drop-down option for Authors. Because each drop-down menu must load every selection for that field, clicking on the drop-down arrow for Authors will most likely stop your computer in its tracks, as we have almost 50,000 names (and growing) in the database.
Drop-down menus allow you to select for a specific item, which can narrow the responses. For example, typing “Biochemistry” in the Journal field will return all journals with “Biochemistry” in the title. However, selecting “Biochemistry” from the drop-down menu will return only entries from the journal “Biochemistry”. Making multiple selections from the drop-down menu will cause the database to list the selections using the Boolean choice “OR”. To change to “AND” or “AND NOT”, simply erase the “OR” (leaving the quotation marks) and enter the preferred option between the quotation marks.
In several fields (e.g., Title, Journals) using free-text entries will prompt the database to return a list of possible matches, which can be selected by clicking on the appropriate choice. The entries returned will then be an exact match to the selection. As the database reads the fields left to right, no suggestions will appear unless the text matches the entry character for character. Otherwise the suggestion field will read “No Matches Found – Use Search for Further Results”. Free text entry means that once the text is entered, using the green Search button will allow the database to search the field for all entries matching the string of characters anywhere in the field.
The Author(s) field is a bit different. The database offers suggestions for this field by last name only. Therefore, for suggestions to be returned, the text must match the last name from the first letter to the last. Adding a space or comma after the last name will stop all suggestions from being listed. To search for John Smith, the text entered into the Author(s) field would be “Smith”, not “Smith,” or “Smith, John”. Otherwise the suggestion field will read “No Matches Found – Use Search for Further Results”. Again, you can then use the green Search button to search the field for all entries matching the string of characters anywhere in the field. However, entries returned may include articles with the consecutive authors “Emily Smith, John Doe”, as the database tries to match the character string.
Dates, PubMed ID numbers, and DOIs can be searched either for the original article or for the relevant notice. One important note about Dates: Some publishers merely overwrite the original article’s HTML page with the retraction notice, making the actual date of retraction impossible to discern. In such cases, the entry will have matching dates for both the original article and for the retraction notice. Some publishers overwrite the original article’s HTML page, but assign the retracted article to an actual print issue. In such cases, the date of the original article is the earlier of the two dates on the HTML page. As resources allow, we are confirming the accuracy of dates but data analyses based on dates alone may have inherent errors.
“Nature of Notice” and “Paywalled” do not offer the free-text option, but instead have three selections apiece: “Retraction”, “Correction” and “Expression of Concern” for Nature of Notice, and “Yes”, “No” and “Unknown” for Paywalled. Leaving any of the fields blank will return all entries for all dates, identification numbers, Notice type and paywall status.
The URL field applies to those entries with a related blog posting from Retraction Watch. To see all the entries with related posts, just type in http in the URL field and click on the green search button. Only the latest 600 entries will be returned, but there are over 11000 (and still counting). You can narrow the search but cut and pasting the specific post’s URL and entering it into the field, except that the “https” string must be changed to only “http”.
More information as to searching each of the fields can be found in Appendix A.
The searches are not case-sensitive, but make sure that you have the spelling used on the publication. You can also check using the wildcard function, i.e. if the author’s name is John Doe try *doe*, in case the name on the article is slightly different, e.g. “John P Doe Jr”. Of course, it may not be the same person, so you should cross-check between fields to be sure.
The fields for Title, Affiliation and Notes are “free text” entry. Searching is best performed by using keywords instead of full titles or affiliations, and use the wildcard character to broaden the search parameters.
The database does not recognize Greek characters, so you may need to spell out symbols such as β as “beta” – or use the wildcard. Diacritical marks are allowed, but are not required for search results, e.g., “Zidek” entered as an author search will return entries with “Zidek” and “Žídek”.
The Affiliation field is a free text field. You can search by using the name of the institution, or any part of the address. As many affiliations are listed differently on different publications (e.g., Bluebird Teaching Hospital might also be listed as Bluebird Teaching and Research Hospital in a separate publication), free text was determined to be the most reasonable way to provide results. A tip: Consider using zip codes or postal codes.
The Reason(s) for Retractions were originally drawn from the tags used in our blog posts, but choices have grown over time. As a general rule, the reasons are selected from the information provided in the notice, coupled with information we have gained through our investigations.
For specific information for each reason for retraction, see Appendix B.
For the most part, the article types can be taken at face value, e.g.,; “Book Chapter/Reference Work” would apply to a chapter in a book, or an entry in an encyclopedic volume.
Most entries are of the type labeled “Research Article,” which we applied to any paper that offered a hypothesis and then explained how and why it was proved or not. The second most common would be “Review Article”; this label was applied to articles that discuss a topic without offering new research findings. However, it also incorporates more specific types of review papers, such as book reviews.
At times, more than a single “Article Type” might be applied to an entry. For example, a research project that was published in the journal’s correspondence section would have both “Letter” and “Research Article” applied to it. A literature review published online ahead of print and thus not assigned to a journal volume would have both “Article In Press” and “Review Article” applied.
Some retractions are taken from different types of sources, such as newspapers, magazines or book chapters. Retraction studies generally focus only on scholarly literature; users should be attentive to the types of articles returned in the searches.
For specific information for each article type, see Appendix C.
Not at all. On the right side of the Search page are two boxes – one labeled “Original Paper” and one labeled “Retraction or Other Notices”. “Original Paper” refers to a title of a published work for which we have entered some type of notice. The “Retraction or Other Notices ” refers to retraction notices, as well as any corrections or expression of concern. You can leave “Nature of Notice” blank to return all relevant notices, or use the dropdown section to select specific types.
Each box on the right hand side allow for searches specific to either articles that have been retracted or specific to retraction notices. Using a MM/DD/YYYY format, you can search within a specific timeframe for the article, the notice or both. Use the “From Date” or “To” fields to give more general time frames (e.g. from 1/1/2000 to present). Some journals do not use a standard date format. To compensate, we are using a date convention for entries:
You can also search using the Pubmed ID (PMID) or by the DOI for the article and/or the associated notice in the boxes on the right side of the Search screen. The wildcard * can be used for partial numbers, e.g. entering “ 28*” in the Pubmed ID field will return all entries with PMIDs beginning with “28”.
You can also search for retraction notices by paywall status – in other words, whether you need a paid journal subscription to view the notice. Free registrations required to view journal contents, where you need to sign into the journal website but need not pay any money to view contents, are not considered “paid” journal subscriptions.
For our purposes, if an entire article (or letter, or review, etc.) was indicated as having been removed from publication either temporarily or permanently, it is considered to be “retracted”. Using words like “Removed” or “Withdrawn” may have procedural meaning to the publisher or journal editor, but the end result is that the published item is not considered appropriate for citation at that time, and we therefore just use the term retracted. We use “Reason(s) for Retraction” to note if the item was removed for a brief period and returned intact (“Temporary Removal”) or if it was republished after changes made (“Retract and Replace”).
Similarly, many publishers and journals have different internal requirements for their use of erratum, corrigendum, and correction. Fundamentally, each means that a change has been made to a published article, but that the article has not been removed from publication. As there is no consistency in use, we just lump all into one category of “Correction”. “Partial Retractions”, where certain sections or conclusions of an article are deemed invalid and not to be used for citation purposes, are also considered “Corrections” as the article remains in journal contents and is still considered citable as a whole.
At this time “Expression of Concern” is being used to cover any editorial note that is not a retraction or correction which addresses the status or reception of a published item.
Often retracted articles have corrections and editorial concerns published prior to (and sometimes after) the official retraction, so we include them as we find them. We also include corrections and expressions of concern when we have posted about them on Retraction Watch, or when they are authored by a person we have posted about.
While we have no plans at this time to include all types of notices in the database, it has not been precluded as a possibility for a later project.
So do we; we are planning that (and a lot of other useful changes) when and if we acquire the funding to do so.
There is an option to acquire a spreadsheet of all the retractions entered into the database, which allows users to then sort and search the information as your desire (up to the date and time of the download). To explore that option, contact team@retractionwatch.com.
Input from users is vital to ensure the best possible outcome, so please feel free to send any comments or questions to team@retractionwatch.com.
If you know of a retraction that you cannot find in the database, you can submit the information through this link. Remember, this isn’t for papers that you think should be retracted; you can send information about those to team@retractionwatch.com.
And thank you for being a part of our work. Your interest and support are what makes this possible.
