The other day I was finishing off supplementary material for an accepted article, and had a major panic for half an hour. It all started with my adding a simple frequency table of the outcome variable: a binary variable. When I checked the PDF it turned out that I have miscoded the outcome variable (at least this is what it looked like) — instead of 60% 1s, I had 40% 1s. What to do? No, I didn’t think the substantive results would have been completely different, so I could have done major work on the page proof, replacing every number in the paper. For a moment I considered ‘unseeing’ what I discovered and bet on the likely case that nobody ever would replicate my findings despite my making all the code and data available. I could even have removed that line where I promise the replication code during the page proof. Ethically defensible this would not have been. Retraction passed my mind. Fortunately, it turned out that there was a benign reason. After going back to and quadruple checking the questionnaire, the raw data, and all the recoding and code, it turned out that I simply wrongly labelled that table of the outcome variable. Relief and feeling silly for panicking.