Allegations of image duplication or manipulation

Journals may apply image integrity checks at any stage in the publication process (submission, before publication or after publication). ... Image integrity screening cannot be expected to uncover all image aberrations, especially in cases with severe misconduct based on an intent to obfuscate information. Conversely, not all image anomalies identified in image screening imply intent to deceive. (Source: STM Working Group recommendations(S’ouvre dans une nouvelle fenêtre) – A2: (1) Scope of journal integrity assessments)

Image manipulation may be detected using the software packages that have been employed in the manipulation (e.g. Photoshop). Tools for detecting image manipulation (termed ‘Forensic Droplets’) are available from the Office of Research Integrity website: http://ori.hhs.gov/droplets(S’ouvre dans une nouvelle fenêtre).

As part of its mission to provide editors with the best available tools to establish the trustworthiness of submissions, Elsevier has been actively seeking image-checking software for several years. Some potential tools have now emerged, which can detect duplication and manipulation in several image types. If you are interested in piloting image checking software, please reach out to your publishing contact. When applying such tools, human oversight is essential to contextualise potential issues flagged in automated software, and discount false positives.

We recommend using a combination of approaches to maximise the chances of detecting manipulated images i.e., reduce false negatives.

The first step should always be to validate the legitimacy of the complaint. The tools described above are intended to support editors with this.

For pre-publication allegations, editors should inform the corresponding authors (or all authors, at editors’ discretion) about their concerns confidentially with the aim of resolving image integrity issues and give them an opportunity to respond (see Form letter A4). In the communication to the corresponding author, the editor should indicate that the matter will likely be referred to the institution or company where the research took place or any other relevant institution or agency (for example a funding agency) unless the author provides a reasonable explanation (accepted as reasonable by the editor).

For post-publication allegations, the complainant must be made aware that the matter cannot be investigated unless the journal editor informs the corresponding author (due process) and the institution or company at which the research took place. Note, additional guidance is provided in the recommendations for comments from third parties e.g., whistleblowers.

At this step, editors may request access to source data or replicate data, where applicable.

Editors may consider publishing an expression of concern when an investigation is underway but a judgement will not be available for a considerable time. See: Corrections to the record (elsevier.com).

Read more: STM Working Group recommendations(S’ouvre dans une nouvelle fenêtre) – A2: (2) Editor responsibilities and transparency; (3) Source data; (4) Replicate data; (7) Considering comments from third parties (readers, whistleblowers).

STM Working Group recommendations(S’ouvre dans une nouvelle fenêtre) - Table 1, provides three levels of classification, based on: the type and severity of the aberrations, any confirmation of image processing error, any evidence of intent to manipulate, and the impact of the affected images to the main conclusions of the research paper. The consequences of each classification level are notably different and vary depending on the explanation provided by the authors and the stage of publication (whether pre- or post-publication).

The editor should consider whether the author’s explanation is reasonable. If applicable, the editor would also normally inform the complainant of the author’s explanation and seek comment (see Form letter B).

The editor should also consider referring the matter to the institution or company employing the author/co-authors and/or the funding agency, noting the allegation of image manipulation (see Form letter C).

Then the editor should refer the matter to the institution or company employing the author/co-authors noting the allegation of image manipulation (see Form letter C) and/or the funding agency (using Form letter E).

Note: To avoid unintentionally aiding authors in destroying evidence of malpractice, Editors should contact institutions at the same time as/before they contact authors “in exceptional cases when journals have evidence of substantive or significant falsification or fabrication of data.”

Read more: STM Working Group recommendations(S’ouvre dans une nouvelle fenêtre) – A2: (2) Editor responsibilities and transparency, (5) Interactions between journals and institutions.

Then the editor should inform the corresponding author and complainant that the editor will seriously consider the judgment of the institution.

Read more: STM Working Group recommendations(S’ouvre dans une nouvelle fenêtre) – A2: (5) Interactions between journals and institutions.

In cases of delay or unavailability of the results of an investigation, editors should work independently on the strength of their findings to take appropriate action to timely protect the integrity of the scholarly record. The editor will be expected to make a determination, in their reasonable judgment, as to the underlying facts and to make a recommendation to Elsevier (and possibly the society for a society journal).

Read more: STM Working Group recommendations(S’ouvre dans une nouvelle fenêtre) – A2: (5) Interactions between journals and institutions.