As part of his job as fraud detector at biomedical publisher Spandidos, John Chesebro trawls through research papers, scrutinising near identical images of cells. For him, the tricks used by “paper mills” — the outfits paid to fabricate scientific studies — have become wearily familiar.
They range from clear duplication — the same images of cell cultures on microscope slides copied across numerous, unrelated studies — to more subtle tinkering. Sometimes an image is rotated “to try to trick you to think it’s different”, Chesebro says. “At times you can detect where parts of an image were digitally manipulated to add or remove cells or other features to make the data look like the results you are expecting in the hypothesis.” He estimates he rejects 5 to 10 per cent of papers because of fraudulent data or ethical issues.
