What Happens When Papers Are Retracted: Understanding Retractions in Science

Why Papers Get Retracted

Retractions happen for a spectrum of reasons, from honest mistakes to deliberate misconduct. Understanding the reason matters, because a retraction for a minor computational error carries very different implications than a retraction for data fabrication.

Honest errors include mistakes in data analysis, mislabeled samples, software bugs that produced incorrect results, or contaminated materials that invalidated experiments. These errors are not the result of misconduct but are serious enough that the conclusions of the paper cannot be trusted. Many researchers who retract papers for honest errors continue to have productive, respected careers.

Irreproducibility occurs when other researchers cannot replicate the findings using the same or similar methods. If multiple independent labs attempt to reproduce a study and consistently fail, the original findings become suspect. Irreproducibility alone does not prove fraud, because the failure could result from subtle differences in experimental conditions, but it does undermine the reliability of the original results.

Plagiarism and duplicate publication involve presenting someone else's work as your own or publishing the same data in multiple journals without disclosure. These are violations of publishing ethics that warrant retraction, though they do not necessarily mean the underlying research is flawed.

Data fabrication and falsification are the most serious forms of research misconduct. Fabrication means inventing data entirely, while falsification means manipulating data, images, or results to produce a desired outcome. These are deliberate acts of deception that can have far-reaching consequences, particularly in fields like medicine where other researchers or clinicians may base decisions on the fraudulent findings.

Ethical violations include conducting research without proper informed consent, failing to obtain institutional review board approval, or violating animal welfare regulations. Even if the science itself is technically sound, ethical violations can warrant retraction because the research was conducted improperly.

The Retraction Process

Retractions can be initiated by the authors themselves, by the journal editor, or by the authors' institution following an investigation. When authors discover their own errors, a voluntary retraction is considered the most responsible course of action. When concerns are raised by other researchers, the journal editor may conduct an investigation or request one from the authors' institution.

The retraction process can be slow. Investigations into research misconduct may take months or years, during which the questionable paper remains in the literature, potentially being cited by other researchers. Even after a retraction notice is published, the original paper often remains accessible online (marked with a retraction notice) because removing it entirely would compromise the scientific record.

A retraction notice is published in the same journal, explaining the reason for the retraction and who initiated it. Well-written retraction notices clearly explain what was wrong and whether any conclusions from the paper remain valid. Unfortunately, many retraction notices are vague or uninformative, using phrases like "the authors wish to retract this paper" without explaining why.

Expressions of concern are a milder measure that journals use when problems have been identified but not yet fully resolved. An expression of concern alerts readers that the paper's reliability is in question while an investigation is ongoing. Some expressions of concern are later upgraded to full retractions, while others are resolved without retraction.

How to Check for Retractions

Retraction Watch is the most comprehensive resource for tracking retractions. This independent website, founded by journalists Ivan Oransky and Adam Marcus, maintains a database of retracted papers with detailed explanations of why each paper was retracted. The Retraction Watch database is searchable by author, journal, subject, and reason for retraction.

PubMed marks retracted papers with a "Retracted Publication" label in search results and on the paper's record page. If you are reading a paper that cites a retracted study, PubMed will also flag the citing paper with a notice. The PubMed retraction filter allows you to specifically search for retracted papers or exclude them from your results.

Most major publishers also mark retracted papers on their websites, typically with a prominent watermark across the PDF and a linked retraction notice. However, these markings are not always consistent or easy to spot, and some publishers are slower than others to update their records.

The Scale of the Problem

The number of retractions has increased dramatically over the past two decades, from a few dozen per year in the early 2000s to over 10,000 in recent years. This increase partly reflects growing vigilance, improved fraud detection tools, and greater willingness to retract problematic papers. Image manipulation detection software, statistical anomaly screening, and crowd-sourced integrity checking on platforms like PubPeer have all contributed to identifying papers that need retraction.

Despite the increase, retractions still represent a tiny fraction of the millions of papers published each year. Some researchers argue that the true number of flawed papers is much higher than the retraction count suggests, because many problematic papers are never formally retracted. Others point out that the increasing retraction rate is a sign of health, not disease, because it means the scientific community is doing a better job of identifying and correcting errors.

High-Profile Retractions and Their Consequences

Some retractions have had far-reaching consequences beyond the retracted paper itself. The 1998 Lancet paper by Andrew Wakefield linking the MMR vaccine to autism was retracted in 2010 after investigation revealed ethical violations and data manipulation. Despite the retraction, the paper contributed to a global decline in vaccination rates and a resurgence of measles that continues decades later. This case illustrates how a single fraudulent paper can cause lasting damage to public health, and why detecting and retracting problematic papers as quickly as possible matters.

In 2014, a paper in Nature describing a simple method for creating stem cells (called STAP cells) was retracted after multiple laboratories failed to reproduce the results and investigation revealed image manipulation. The case drew international attention, led to the resignation of senior researchers, and prompted institutional reforms in data management and research oversight. More recently, concerns about image manipulation and data integrity have led to investigations of thousands of papers across multiple fields, with AI-powered tools detecting suspicious patterns that human reviewers missed.

These cases demonstrate that retractions are not merely academic housekeeping. They can affect clinical practice, public policy, investment decisions, and public trust in science. The speed and transparency of the retraction process directly influences how much damage a flawed paper can cause before it is corrected.

What Retractions Mean for Readers

A retracted paper should not be cited as evidence. If you discover that a paper you have been relying on has been retracted, check the retraction notice to understand why, then look for alternative evidence for the claim the paper supported. If other independent studies have found similar results, the retracted paper may not change the overall picture. If the retracted paper was the sole or primary evidence for a claim, that claim is now unsupported.

Retracted papers continue to be cited, sometimes for years after retraction. Studies have found that many researchers cite retracted papers without noting the retraction, either because they are unaware of it or because they accessed an older cached version. Making a habit of checking for retractions, especially for papers that are central to your conclusions, is a worthwhile practice.

When a retracted paper was part of a larger body of work, consider the implications for related papers by the same authors. A retraction for data fabrication raises questions about the integrity of the researcher's other publications, while a retraction for an honest error in one study does not necessarily cast doubt on unrelated work. Institutional investigations sometimes examine an author's full publication record when serious misconduct is found, and additional retractions may follow. Retraction Watch tracks these cascading investigations and provides context about whether concerns extend beyond a single paper.