How exam questions are written

Every PMP question is written by trained item writers, reviewed by other practitioners, tested as an unscored pretest item and analysed statistically before it becomes a scored question. That process produces questions with a recognizable structure. Once you can see the structure, you read faster, you notice the words that decide the answer, and you are far less likely to fall for options that are attractive but wrong. This lesson takes a question apart, then looks at how the different formats work and how wrong options are built.

The questions below are written in the same style as the exam, using Project Compass. They are practice items written for this course, designed to teach the patterns you will meet.

The anatomy of a situational question

Most PMP questions are situational. They describe a project, a moment in its life and a problem, then ask what the project manager should do or what is most likely true. A typical question has four parts.

Consider this question. "Project Compass is using a hybrid approach. Midway through the fit-out, the electrical contractor tells the project manager that a shortage of a type of breaker panel will delay delivery by three weeks. The panels are on the critical path. What should the project manager do first?" The context is hybrid, but this part of the work is predictive construction. The situation is that a risk has become a real problem on the critical path. The stem asks what to do first. A good answer would involve assessing the impact on the schedule and looking at options, such as alternative suppliers or resequencing, before escalating or changing the baseline.

Each part narrows the answer. If the situation said the shortage might happen, the answer would involve risk responses. If the stem asked what should have been done, the answer would point to earlier planning, such as identifying long-lead items. If the context were an agile software team, the answer would probably involve the team and product owner in reordering work. Reading all four parts before looking at the options is the single habit that does most to improve accuracy.

The words that decide the answer

Item writers use a small set of words to signal what kind of answer they want. Read them as instructions.

The word first deserves special attention. When several options are good actions, the question is about order. Gathering information and assessing impact usually come before acting, acting within your authority usually comes before escalating, and talking to the person involved usually comes before going to their manager. When a question says the project manager has already reviewed the impact, the first remaining step moves on to the next stage, such as discussing options with the team.

Single and multiple response

In single-response questions, choose the one best option. In multiple-response questions, the stem says how many to choose, for example: "Which two actions should the project manager take?" Preparation providers generally report no partial credit for multiple-response items, so every choice must be right. Treat each option as a true or false statement about the situation, keep the ones that are clearly true and compare the remaining candidates carefully.

# Why eliminating wrong options matters: expected marks from 20 questions you are unsure about.
# Never leave a question blank; there is no penalty for a wrong answer.
unsure = 20
for left in (4, 3, 2):
    expected = unsure / left
    print(f"options left {left}: expected correct {expected:4.1f} of {unsure}  ({1 / left:.0%} each)")
print()
# Multiple response "choose two of five": a blind guess is rarely right.
from math import comb
print(f"choose 2 of 5 blind: {1 / comb(5, 2):.0%} chance")
print(f"choose 2 of 5 after ruling out 2: {1 / comb(3, 2):.0%} chance")
print(f"choose 2 of 5 when sure of 1, guessing the other from 3: {1 / 3:.0%} chance")

options left 4: expected correct  5.0 of 20  (25% each)
options left 3: expected correct  6.7 of 20  (33% each)
options left 2: expected correct 10.0 of 20  (50% each)

choose 2 of 5 blind: 10% chance
choose 2 of 5 after ruling out 2: 33% chance
choose 2 of 5 when sure of 1, guessing the other from 3: 33% chance

The numbers show why elimination matters. If you can rule out two of four options, a forced choice between the last two is right half the time, compared with a quarter when guessing blind. There is no penalty for wrong answers, so never leave a question unanswered. For a choose-two-of-five question, a blind guess succeeds only one time in ten, but ruling out two clearly wrong options raises that to one in three.

Case or scenario sets

A case set gives a longer scenario and then several questions about it. The scenario may include a stakeholder list, a status report, a burndown chart or a short email thread. The questions may test different domains: one might ask about a conflict, another about the schedule, another about a compliance requirement. On the 2026 exam the first break comes after the case-study section, and once a section is closed you cannot return to it, so finish and review the case questions before you take the break.

An efficient routine for case sets is:

• Skim the questions first, so you know what to look for.
• Read the scenario once, carefully, noting the approach, the stage, the key people and any numbers.
• Answer each question from the facts in the scenario, not from assumptions. If the scenario says the team is self-organizing, that fact governs every question in the set.
• Watch for questions that add new information. A later question may say "Two weeks later, the sponsor resigns." That fact applies to that question only, unless it says otherwise.

Graphic-based questions

Graphic-based questions show a chart, task board, network diagram, histogram or table. The commonest errors come from reading the graphic too quickly. Before reading the question, read the title, the axes and their units, the legend and any notes. On a burndown chart, check whether the line shows work remaining or work completed; on a cumulative flow diagram, check which band is which state; on a network diagram, check whether durations are in days or weeks. Then read the question and return to the graphic for exactly the value you need.

How wrong options are built

Wrong options, called distractors, are written to be plausible. Knowing the common types helps you see through them.

Many wrong options are not bad actions in general; they are wrong for this situation. Updating a baseline is sometimes right, but only after an approved change. Escalating is sometimes right, but only when the problem is beyond your authority or attempts to solve it have failed. That is why reading the situation carefully matters more than memorizing rules such as never escalate.

Questions with incomplete information

Exam scenarios are short, so they leave things out. A common mistake is to fill the gaps with assumptions from your own workplace. If the question does not say that a process exists, do not assume it does not; if it does not mention a contract, do not invent one. Answer from the facts given and from the practices described in PMI's standards. When a key fact really is missing, the best answer is often the one that gathers it, such as reviewing the risk register or meeting the stakeholder.

This matters for Canadian candidates in a particular way. Many questions describe practices in general terms, without the provincial rules you may know from work, such as holdback under a provincial construction act or a provincial privacy law. Answer the question as asked; if a law is relevant, the question will say so.