How these questions are written and checked
This page exists because “clinically reviewed” is easy to type and hard to verify. Here is what actually happens to a question before it appears on this site — and what still gets past us.
- Written
- 156
- Published
- 130
- Discarded
- 17%
The process
- 1
Write to the blueprint, not from the exam
Items are written from standard nursing knowledge against the exam’s published subject areas. Nothing here is copied, recalled, or reconstructed from a real exam. Reproducing live exam content is prohibited by the test provider, and buying “real questions” is what gets candidates barred.
- 2
Hand it to a reviewer told to refute it
Each item goes to an independent clinical audit whose instruction is not “approve this” but “prove this wrong.” It re-derives the answer, recomputes any arithmetic, checks each option’s rationale separately, and looks for ambiguity and answer cues.
- 3
Rewrite, then audit again
Flagged items are rewritten to address every specific objection and sent back through the same audit. An item has two chances.
- 4
Publish only what passes
Anything still flagged is discarded rather than shipped with a caveat. Near-duplicates and items where the correct answer is guessable by length are also dropped at this stage.
What gets a question rejected
The keyed answer is wrong
The most serious failure. A reviewer re-derives the answer independently; if the key does not hold up against current practice, the question is rewritten or dropped.
The guidance is out of date
Practice moves. Items keyed to superseded guidance get caught here — for example, routine gastric residual monitoring, or giving oxygen to a normoxic patient with acute MI.
A rationale is false
A correct answer with a wrong explanation still teaches the wrong thing, so every option's rationale is checked on its own, not just the keyed one.
A calculation does not check out
Every dose and drip-rate item is recomputed from the numbers in the stem. A doubled heparin rate is exactly the kind of error this stage exists to catch.
More than one answer is defensible
If a competent nurse can argue for a second option, the item is not single-best and gets reworked or removed.
The answer is guessable without knowing the content
Longest option, odd grammar, or the only option in its category — all of these let a test-wise candidate score without knowing the material.
What this process does not do
Being straight about the limits, because a study aid that oversells itself is not worth trusting:
- It is AI-assisted. Questions are drafted and audited with language models. That catches a great deal — wrong keys, stale guidance, broken arithmetic — but a fact that both the writer and the reviewer have wrong can survive. Independent review by a licensed nurse is the backstop we are adding as the bank grows.
- It is practice, not prediction. These questions are built to the same subject areas as the exam. They are not the exam, and no one can honestly tell you what will be on yours.
- Guidelines change. An item correct today can be superseded. If you spot something that has moved on, tell us and we will fix it.
Found an error? That is genuinely useful — it is how the bank gets better. Note the question id and what you think is wrong, and send it to corrections@rationalebank.com.