Discussion about this post

User's avatar
Overcoming Bias Commenter's avatar

Anna: Not a checklist, but I had the idea to take one bias that seems problematic at a time, write a note of it somewhere prominent, and try to keep it in my mind and watch out just for it, in the hope that I come to (sometimes) catch it automatically. It's too early to say how it works.

Expand full comment
Overcoming Bias Commenter's avatar

Hmmm... What if you had a clinical trial that was less discrete and more continuous?

For example, you could have every doctor participating in the trial get a different subset of the 50 items you are considering for your checklist. This would allow you to do things like write patient health as a function of checklist length and write out the checklist items in order from least effective to most effective.

Instead of testing the safety function at a few points (e.g. giving every doctor a 12-item checklist), you could test the safety function at a wide variety of points (6, 12, 18, 24, 30, and 36-item checklists) and then write a curve to match your data.

What are the disadvantages of this approach?

Expand full comment
7 more comments...

No posts