@ tavie
Yes, it was a data entry error. The whole table is inconsistent. A fair amount went wrong in that post. The problem is that the cross-check only verified the raw findings, not the percentages.
It's easy to lose track in the volume of numbers, and when youfinally write it up - building the tables, writing the text - every cell has to be verified.
Honestly, this whole episode has me asking whether posting work-in-progress findings is worth it when errors like this can still slip through despite reasonable care. That raises the question of the cost-benefit ratio. I'll think about it.
I thought sharing the results might help others, maybe spark ideas. That was also the point of the "Work in Progress"framing. But mistakes like this happen, and the responsibility for catching them before publication is mine.
Yes, it was a data entry error. The whole table is inconsistent. A fair amount went wrong in that post. The problem is that the cross-check only verified the raw findings, not the percentages.
It's easy to lose track in the volume of numbers, and when youfinally write it up - building the tables, writing the text - every cell has to be verified.
Honestly, this whole episode has me asking whether posting work-in-progress findings is worth it when errors like this can still slip through despite reasonable care. That raises the question of the cost-benefit ratio. I'll think about it.
I thought sharing the results might help others, maybe spark ideas. That was also the point of the "Work in Progress"framing. But mistakes like this happen, and the responsibility for catching them before publication is mine.