note
One Rate Decision, Four Rejection Labels: Why a Single Count Can't Rank Them
A reader could ask which rejection label fires most across research scans. The supplied rate-decision item carries no rejection counts, scan totals, or limitation labels, so it cannot rank them. I explain why a rejection tally is roughly a rejection rate times a logging rate, and what per-reason, per-scan counts would be needed before any label could be called binding.
What the source carries, and what it does not
Observation: the supplied public-news-feed item reports that rates were hiked in a unanimous decision despite fierce opposition from a president who had called for a cut, and it carries no rejection counts, no scan totals, and no limitation labels. On the comparison the reader wanted — sample size versus positive rate versus net capture versus data quality — that tally ranks nothing, because there is no per-label count in it to compare.
My view, first person and low confidence: a rejection tally behaves roughly like a rejection rate multiplied by a logging rate, so one total is compatible with candidates failing on any of the four limits. Change how often scans run or how often a reason gets written down, and the same headline appears under a different binding constraint. That arithmetic is my interpretation, not something the source asserts; the source is a rate decision, not a validation claim, and I am not reading it as one.
What would change the ranking
Hypothetical: four labels, one reason recorded per rejection, counted per scan rather than summed into a grand total. Falsified if most rejections carry no reason, or if a label leads only because it is the cheapest one to log. The supplied sample-size guide makes the related trading point — a trade count is not an evidence count, since rows from one signal in one regime can carry less independent information than fewer rows spread across conditions — and the same coverage objection applies to rejection rows.