September 8, 2026 · 8 min read
TL;DR: We split every answer option by what kind of noun it names. Documents: 24.6% keyed. People: 26.6% keyed. Neither: 24.6% keyed. Baseline is 25%. Naming an artefact or a stakeholder carries no information whatsoever about correctness, which makes this the most useful thing we have measured that turned out to be nothing.
A lot of exam advice is built on nouns. Look for the risk register. The answer usually involves the team. Be suspicious of options that name a document.
We tested all of it. None of it holds.
Every option across all 1,948 single-answer questions was classified by the nouns it contains, then split by keyed versus distractor.
| Option contains | Keyed | Distractor | Keyed rate |
|---|---|---|---|
| A project document (register, log, plan, charter, baseline, matrix, backlog, report) | 271 | 829 | 24.6% |
| A person or group (team, sponsor, stakeholder, customer, owner, vendor, board) | 605 | 1,667 | 26.6% |
| Neither | 1,159 | 3,552 | 24.6% |
Baseline is 25%. All three land on it. The largest gap in the table is 2 percentage points, which on these counts is not worth acting on.
Break the documents out one by one and the picture holds:
| Document named in the option | Keyed | Distractor | Keyed rate |
|---|---|---|---|
| Change log | 3 | 5 | 38% |
| Stakeholder register | 7 | 15 | 32% |
| Backlog | 47 | 106 | 31% |
| Issue log | 12 | 29 | 29% |
| Work breakdown structure | 14 | 34 | 29% |
| Project charter | 15 | 39 | 28% |
| Lessons learned | 12 | 36 | 25% |
| Risk register | 16 | 49 | 25% |
| RACI | 1 | 6 | 14% |
| Assumption log | 0 | 3 | 0% |
Not one of the well-populated rows breaks meaningfully from baseline. The risk register, the single most-cited artefact in PMP folklore, lands on exactly 25%.
Availability. When "refer to the risk register" is right, it is right for a satisfying reason, and the reason makes a story you remember.
Take the opening question of this bank. A significant quality problem surfaces with a key supplier's deliverables. Buried in the stem: this particular supplier risk had been identified earlier in the project.
The register wins because the scenario said the risk was already identified, which means a response was already agreed. Brainstorming a solution that exists is wasted effort. It is a clean, teachable moment, and it lodges.
The 49 questions where the register is a wrong answer do not lodge, because they are unremarkable. The register is offered, the risk was never identified, and consulting it would find an empty page. Same noun, opposite verdict, and nothing memorable to attach it to.
Common trap: "The correct answer usually points at a project artefact." Across this bank, artefact-naming options are keyed 271 times and wrong 829 times, dead on the 25% you would get by guessing. What separates the 271 from the 829 is never the artefact. It is whether the scenario established that the artefact already contains what you need. This bank states which sentence in the stem made the register the right place to look, rather than just marking the option wrong.
The verb and the timing, which we have measured and which do carry signal.
| Signal | Keyed rate | Baseline |
|---|---|---|
| Option opens with Analyze | 83% | 25% |
| Option is one of a matched pair of twins | 65% | 50% |
| Option contains sequencing before | 38% | 25% |
| Any option naming a document | 24.6% | 25% |
| Option contains so that | 8.6% | 25% |
| Option opens with Report or Continue | 2% | 25% |
Structure carries information. Vocabulary mostly does not. That is worth internalising as a general posture toward exam-technique advice, because noun-based rules are the easiest kind to invent, the easiest to remember, and the hardest to check.
The opposite, and this is where the null result becomes useful rather than merely deflating.
Because the noun is a coin flip, you cannot shortcut the question. You have to know what each artefact is actually for, so you can tell whether this scenario is one where it helps:
Each one has a job. Reading the artefact list as a set of triggers, "who is missing" for the stakeholder register, "was this anticipated" for the risk register, is what replaces the noun rule that does not work. It is more to learn, and it is the only version that survives measurement.
Are PMP answers that mention a project document usually correct? No. In this bank, options naming a register, log, plan, charter, baseline, matrix or backlog are keyed 271 times against 829 distractors, which is 24.6 percent against a 25 percent baseline. Naming an artefact is worth nothing on its own.
What about options that name the team or a stakeholder? Also nothing. Options naming a person or group are keyed 605 times against 1,667 distractors, 26.6 percent. Statistically indistinguishable from options naming a document, and from options naming neither, which sit at 24.6 percent.
Then why does the risk register feel like a correct answer so often? Because when it is correct, the explanation is memorable: the risk was already identified, so the response is already written down. That is a vivid pattern and it sticks. The 49 times it appears in a wrong answer are forgettable by comparison.
What actually decides whether an artefact answer is right? Whether the scenario established that the artefact contains what you need. Refer to the risk register is right when the stem says the risk was identified earlier and wrong when it was not. The verb and the timing carry the signal; the noun never does.
PMP Practice's 2,141 questions are re-certified against PMBOK 8 and the July 2026 ECO, with every wrong answer explained, not just marked wrong, including why the artefact it named was the wrong place to look. Start the free 20-question sample - no card, no signup required to try it.
Are PMP answers that mention a project document usually correct?
No. In this bank, options naming a register, log, plan, charter, baseline, matrix or backlog are keyed 271 times against 829 distractors, which is 24.6 percent against a 25 percent baseline. Naming an artefact is worth nothing on its own.
What about options that name the team or a stakeholder?
Also nothing. Options naming a person or group are keyed 605 times against 1,667 distractors, 26.6 percent. Statistically indistinguishable from options naming a document, and from options naming neither, which sit at 24.6 percent.
Then why does the risk register feel like a correct answer so often?
Because when it is correct, the explanation is memorable: the risk was already identified, so the response is already written down. That is a vivid pattern and it sticks. The 49 times it appears in a wrong answer are forgettable by comparison.
What actually decides whether an artefact answer is right?
Whether the scenario established that the artefact contains what you need. Refer to the risk register is right when the stem says the risk was identified earlier and wrong when it was not. The verb and the timing carry the signal; the noun never does.