I also didn't think this. I've read the book but not the article, and while the conclusion seems fine, a lot of the material in the book is stretched.
The one that bothered me the most was the inclusion of the India soap program as support for checklists -- those are entirely different models.
The soap model involved researchers handing out free soap, and teaching people that they should wash with the soap in any of a set of specific circumstances:
- once a day (full body)
- before preparing food (hands)
- before eating (hands)
- before distributing food to anyone (hands)
- after defecating (hands)
- after wiping an infant (hands)
This made a big dent in the prevalence of disease in experimental neighborhoods. Great! This was a good idea. But it's not an example of a checklist. The concept here is that, every time you do anything at all, you see whether it's one of those five hands-washing circumstances, and if it is, then you wash your hands. This is the opposite of a checklist, where you perform a series of verifications whenever you take a specific particular action, not whenever you take any action at all.
(Obviously, people can't handle the mental load of "before doing _anything_, check to see..." and instead would have added "wash your hands" to the appropriate five behavioral sequences. That reduces the mental load from (1) a constant mental drain on any activity of any kind to (2) learning five things. But if you make that switch, you stop having any relationship to the checklist concept at all.)
