Writing aviation MCQs that test understanding, not memory

Good aviation exam questions test whether a candidate can use what they know, not whether they can repeat it. An item that asks only for a remembered label rewards revision tactics, while one that asks for a decision, a calculation or an interpretation shows whether the training worked. This guide is for chief instructors, exam officers and quality managers at aviation training organizations who write or approve items. It is not for candidates looking for practice papers.

It separates the few item-writing rules that appear in regulation from the many that are only good practice, then shows how to climb from a recall item to an application item on one topic. Every illustrative item here is original, written for this post, and not representative of any official exam.

The regulatory points come from the UK Civil Aviation Authority’s published copy of Regulation (EU) No 1321/2014 and the FAA’s Aviation Instructor’s Handbook, read as of October 2026. Rules differ by country and change over time, so check your competent authority before you change a live exam.

Key takeaways

• Regulation says little about how to write items. The Part-66 basic examination standard is the clearest exception: three alternatives, one correct, all plausible and similar in wording and length (as read in the UK CAA copy, October 2026).
• An item should demand the same level as its learning objective. Part-66 knowledge levels run from familiarization with terms to interpreting measurements, and each level calls for a different item pattern.
• Wrong answers should come from real procedural errors, such as an inverted conversion or swapped readings, not from random numbers or filler.
• Part-66 Appendix II allows a nominal average of 75 seconds per question, so an item that needs minutes of reading or arithmetic does not fit the paper.

What do the rules say about writing aviation exam questions?

Very little. Aviation regulation sets item-writing rules in only a few places. The clearest found is the Part-66 basic examination standard, which fixes three alternatives and requires plausible, similar-length options. Most other advice, such as the FAA’s instructor handbook, is guidance rather than law.

Appendix II of Part-66, titled “Basic Examination Standard (except for category L licence)”, is the source. In the UK CAA’s published copy of Appendix II, paragraph 1.1 says “The incorrect alternatives shall seem equally plausible to anyone ignorant of the subject” and “All of the alternatives shall be clearly related to the question” (quoted). It adds that alternatives should be “of similar vocabulary, grammatical construction and length” (quoted). Paragraph 1.2 requires three alternatives, only one of them correct.

The FAA’s Aviation Instructor’s Handbook, Appendix B gives instructors item-writing advice. It is guidance for instructors, not a rule for any particular test. The table below shows each rule and how binding it is.

Two limits apply. This post read the UK copy, not the EASA Easy Access Rules, so confirm the wording in the version your competent authority uses. And who writes, reviews and approves your question bank is set by your competent authority and your organization’s procedures, which this post has not read.

What is the difference between testing memory and testing understanding?

A memory item asks the candidate to repeat something they were taught. An understanding item asks them to use it in a situation they have not seen. The FAA handbook calls the lowest level of learning “the ability to repeat something which one has been taught” (quoted) and lists rote, understanding, application and correlation as four practical levels.

The handbook’s learning-process chapter sets out those levels. Part-66 gives maintenance training its own ladder. In the UK CAA copy of Appendix I, Level 1 is “A familiarisation with the principal elements of the subject.” Level 2 is “A general knowledge of the theoretical and practical aspects of the subject”. Level 3 is “A detailed knowledge of the theoretical and practical aspects of the subject” (all quoted).

Appendix I also lists what an applicant should be able to do. At Level 1, “use typical terms”; at Level 3, “interpret results from various sources and measurements” (quoted). Those verbs show which item pattern fits. The table below is this post’s own mapping from level descriptions to item patterns, not part of the regulation.

What do the three levels look like on one topic?

Take Ohm’s law. The three illustrative items below test it at three levels, each with three options to match the Part-66 format.

Level 1 item. Which unit is used to measure electrical resistance?

• A. Henry
• B. Ohm (key)
• C. Farad

Level 2 item. A circuit has a constant 28 V supply and one fixed resistor. The resistance is doubled. What happens to the current?

• A. It halves. (key)
• B. It doubles.
• C. It stays the same.

Level 3 item. A technician measures 24 V across a heater element and 3 A through it. Later, at 24 V, the same element draws 2 A. What has happened to its resistance?

• A. It has risen from 8 to 12 ohms. (key)
• B. It has fallen from 12 to 8 ohms.
• C. It has stayed at 8 ohms.

The Level 1 item is a recall item, and that is legitimate when the objective is to use the term. The Level 2 item cannot be answered by naming the law. In the Level 3 item, option B comes from swapping the two readings and option C from assuming the voltage fixes the resistance. Each wrong answer is a mistake someone could really make.

Pro tip: Write the Wrong Answers First. Before you write a stem, list three mistakes your own candidates make: from marked scripts, practical sign-offs or instructor notes. Build the options from that list. A distractor taken from a real slip is plausible to the people it is meant to catch.

How do you rewrite weak aviation exam questions to test understanding?

Name the level the objective needs, replace the label with a situation or measurement, decide what the stem must supply, write wrong answers from known errors, and work the key from the stem alone. Run each draft item through these five steps before it enters the bank.

Step 1: Name the objective’s level. Underline the verb in the learning objective. “Identify” points to a Level 1 pattern, “explain” or “apply” to Level 2, and “interpret” to Level 3. If the verb and the item disagree, change one of them.

Step 2: Replace the label with a situation. Instead of “What is the unit of resistance?”, describe a circuit and ask what changes. The post on scenario-based multiple-choice questions covers the general stem template.

Step 3: Decide what the stem supplies. If the objective is the procedure, give constants and conversion factors in the stem. If the objective is recalling the constant, leave it out. Record the choice in your item specification, or the item fails for the wrong reason.

Step 4: Write wrong answers from known errors. The FAA handbook recommends common learner errors as distractors (table above), and the tip below shows where to find them.

Step 5: Work the key from the stem alone. Ask a second subject-matter expert to answer without seeing your key. If they disagree, the stem is ambiguous or the key is wrong. Fix it before piloting.

How do you write numeric wrong answers from real procedural errors?

Compute each wrong answer by applying a specific mistake to the same inputs as the key. Part-66 Appendix II, paragraph 1.1, says incorrect answers in numerical questions “shall correspond to procedural errors” and “shall not be mere random numbers” (quoted). Each wrong option should trace to one named mistake.

The paragraph’s examples are corrections applied in the wrong sense and incorrect unit conversions. This illustrative item uses a conversion. The torque value is invented and comes from no maintenance manual.

Illustrative numeric item. A task requires 120 lbf·in. Your torque wrench reads in N·m. Use 1 lbf·in = 0.113 N·m. What setting is needed?

• A. 13.6 N·m (key)
• B. 163 N·m
• C. 1,062 N·m

There is no random value here, such as 11.2 N·m, that no candidate would reach by any route. A candidate can discard a random distractor without calculating, which leaves an effective two-option item. With only three options allowed, you cannot afford a weak one.

How long should an aviation exam question take to answer?

Under Part-66 Appendix II, exam time per module is based on a nominal average of 75 seconds per question. That is an average, not a limit per item. A reasonable item takes a prepared candidate about that long, so stems with long passages or multi-step arithmetic need a deliberate decision about time.

Paragraph 1.2 of Appendix II allows “a time per module which is based upon a nominal average of 75 seconds per question” (quoted). As arithmetic, a hypothetical 40-question paper at that average runs 50 minutes (40 × 75 seconds = 3,000 seconds). Module question counts and times are in Section 2 of Appendix II, which this post does not reproduce.

The consequence is that application items cost time. A three-step calculation with a table to read can take several minutes, using the time of several recall items. Balance the paper by time, not only by topic.

Pro tip: Time the Item With a Recent Graduate. Authors and experts read fast because they know the answer. Ask someone who passed the course recently to answer the item cold while you time them. If it takes much more than 75 seconds, shorten the stem or split the item.

What checklist should an aviation MCQ pass before it enters the bank?

An item is ready for the bank when it names its objective and level, can be answered from the stem, has wrong answers tied to known errors, uses parallel options, and has a key confirmed independently. This checklist combines the rules in the first table with the checks above.

1. The item names one learning objective and the level it needs.
2. The stem contains every fact needed to answer and nothing irrelevant.
3. Each wrong answer maps to a named misconception or procedural error.
4. The options match in vocabulary, grammar and length. The FAA handbook notes that “correct alternatives are longer than incorrect ones” in instructor-made tests (quoted), so check yours.
5. The stem avoids double negatives and absolutes such as “never” and “always”, as the FAA handbook advises.
6. Numeric wrong answers are computed from specific errors.
7. A second expert, working only from the source document, reaches the same key.
8. A recent graduate can answer it in roughly the nominal time.
9. The item is original, not copied or adapted from a real exam, a published question bank or a practice-question site.

Tag each item with its objective and syllabus reference so it can be found and retired. The post on building an aviation question bank by ATA chapter covers the bank structure.

Where does memory still belong in aviation exam questions?

Memory belongs wherever the objective says the candidate must know something without reference: a term, a unit, a limit or a procedure step. Recall items are not a flaw. The flaw is using them for an objective that asks for application.

The FAA handbook advises questions “that call for essential knowledge rather than for abstract background knowledge or unimportant facts” (quoted). Test each recall item: could a competent person work safely without knowing this unaided? If not, replace it with an application item and record the mix by level in the blueprint.

How do you check with data that items test understanding?

Item statistics cannot prove an item tests understanding, but they flag items that behave oddly. Review difficulty, discrimination and option counts after each cohort, send flagged items to an expert, and treat small cohorts with caution.

The post on item analysis for trainers shows the calculations and why cutoffs differ between sources. Two patterns matter here. A distractor that strong candidates pick may signal a second defensible answer or a wrong key. A distractor nobody picks does no work, and with three options it leaves a two-option item.

If you draft items with AI tools, expect first drafts to drift toward recall and run the level check above. The review workflow for AI-drafted questions describes the gates an expert should apply.

What should you do next?

Pull 20 items from your bank and label each with the level its objective needs and the level it tests. Rewrite the mismatches with the five steps above, starting with any numeric item whose wrong answers are random. Then ask two experts to answer them blind.

When choosing tooling, ask any vendor how it handles a three-option limit, per-option statistics and tagging by syllabus reference. SimpliAssess’s published features page names item banking and tag management, but it does not list question types or option-count settings, so confirm those in a demo.

On the training-operations side, SimpliTrain’s aviation page says the platform supports delivery and tracking of EASA Part-147 Basic Training, Type Training and OJT completion. It lists Assessments & Exams and Question Banks only as menu labels, so it does not describe item-writing features.

Frequently asked questions

How many options should an aviation multiple-choice question have?

For Part-66 basic examinations, three, with only one correct, according to paragraph 1.2 of Appendix II in the UK CAA copy read in October 2026. The FAA’s Aviation Instructor’s Handbook says three or four alternatives are generally provided. Other exams, such as airline or cabin crew tests, may set their own rules, so check your competent authority.

Can multiple-choice questions test application, not just recall?

Yes. The FAA handbook describes multiple-choice items as able to test from the acquisition of facts to understanding, reasoning and the ability to apply what has been learned. Put the candidate in a situation or give them measurements to interpret, as in the Ohm’s law items above.

What is the Part-66 pass mark for multiple-choice questions?

Paragraph 1.7 of Appendix II, in the UK CAA copy read in October 2026, sets it at 75 percent for each module and sub-module multiple-choice part. Paragraph 1.10 says penalty marking shall not be used to decide whether a candidate has passed. Confirm both against the version your competent authority applies.

How long should candidates get per question?

Part-66 Appendix II bases module time on a nominal average of 75 seconds per question. It is an average used to set module time, not a limit on each item. Multi-step items use more than their share, so balance the paper by time as well as topic.

Is it acceptable to copy questions from practice sites?

No, write original items. Material from practice sites or exam dumps has unknown origin and may be copyrighted. It was also not written against your objectives, so it cannot show whether your training worked. If an item resembles real exam content, raise it with your competent authority.

How do you know a distractor is plausible?

Build it from a known error, then test it. Part-66 requires wrong answers to seem equally plausible to anyone ignorant of the subject. After a sitting, check that some candidates chose each option and that stronger candidates did not favor a wrong one. The item analysis post explains thresholds and limits.