Knowing how to construct a test well is one of the most useful skills a teacher has, and one of the least practised. Most teachers learned test construction in one college course, years ago, and have been making tests under deadline ever since. The result is familiar: tests that are mostly recall, items that two learners read two different ways, and answer keys corrected after the papers come back. None of that is a lack of effort — it is a lack of a process.
This guide is that process. It covers how to construct a test from start to finish in ten steps — purpose, blueprint, item types, writing, review, assembly, answer key, try-out, scoring, and analysis — with the rules that matter at each step and links to deeper guides where a step deserves its own article. It works for a ten-item quiz, a summative test, or a term examination.
The 10-second answer: Decide what the test is for and what it covers, build a table of specifications, choose item types that fit each competency, write and review the items, assemble them into parts with clear directions, prepare an answer key and scoring guide, try the test before using it, then score it and run an item analysis so the next version is better. Good tests come from the blueprint, not from the item you happen to think of first.
What makes a test good
Before the steps, it helps to know what you are aiming for. Assessment specialists judge a test by four qualities, and every step in this guide exists to protect at least one of them.
| Quality | Plain meaning | Protected by |
|---|---|---|
| Validity | It measures what you taught and meant to measure | The blueprint and well-written items |
| Reliability | It would give similar results if given again | Enough items, clear wording, consistent scoring |
| Fairness | No learner is helped or hurt by things unrelated to the skill | Plain language, familiar formats, no trick items |
| Practicality | It can be given and checked in the time available | Sensible length and an efficient answer sheet |
If you want the statistics behind reliability, our guide to KR-20 and point-biserial explains them without the textbook language.
Step 1: Define the purpose and the coverage
Every decision later depends on why the test exists. A diagnostic test at the start of a unit, a quick formative check during it, and a graded summative test at the end look different, because they answer different questions. Write the purpose down in one sentence before anything else.
Then fix the coverage: the exact competencies or learning outcomes the test will measure. Under DepEd Order No. 15, s. 2026, tests are teacher-made and based on what was actually taught, so the coverage comes from your lesson log, not from the full curriculum guide. If you are unsure which assessments are graded, our guide to formative vs summative assessment under DO 15 sets it out.
Step 2: Build the blueprint (table of specifications)
The table of specifications turns the coverage into numbers: how many items each competency gets, in proportion to the time spent teaching it, and how those items spread across cognitive levels. It is the single most effective way to stop a test from drifting toward whatever was easiest to write.
The core calculation is items for a competency = (days on that competency ÷ total days) × total items. Our TOS guide with a DepEd sample walks through it by hand, the Filipino version is Paano Gumawa ng TOS, and if you would rather not do the arithmetic, a table of specifications generator does it for you.
Step 3: Choose the item types
Different item types reveal different kinds of learning. The right choice depends on the competency and the cognitive level, not on habit. Most tests combine a few types grouped into parts.
Selected-response items
In selected-response items, learners choose an answer: multiple choice, true or false, and matching type. They are quick to check and objective to score, and multiple choice in particular can reach Applying and Analyzing when the stem presents a real situation rather than a definition.
Constructed-response items
In constructed-response items, learners produce the answer: completion, short answer, problem solving, and essay. They show reasoning that selected-response items cannot, at the cost of slower and less consistent checking unless a clear scoring guide is used.
| Type | Strength | Limitation |
|---|---|---|
| Multiple choice | Covers a lot quickly; can test higher levels | Hard to write good distractors |
| True or false | Very fast to answer and check | 50% chance of guessing correctly |
| Matching type | Efficient for related facts and terms | Mostly limited to recall |
| Completion / short answer | Recall without cues | Several wordings may be correct |
| Essay / problem solving | Shows reasoning and organisation | Slow to check; needs a rubric |
Step 4: Write the items
Good items share one property: the only way to answer them correctly is to know the thing being tested. Each item type has its own ways of breaking that rule, so the guidance below is grouped by type.
Multiple-choice items
Multiple choice is the workhorse of most tests, and most of its problems come from the options rather than the stem. Write the stem as a complete question, then write wrong options from real misconceptions.
- State the problem fully in the stem; avoid stems like “Photosynthesis…” that only make sense after reading the options.
- Use three or four plausible options; one obviously silly option is a wasted option.
- Keep options similar in length and grammar so none stands out.
- Avoid negatives in the stem; if you must use one, write it in capitals: “Which is NOT…”
- Avoid “all of the above” and “none of the above” as habits.
True-or-false items
True-or-false items look easy to write and are easy to write badly. Each statement must be entirely true or clearly false, with no qualifiers that make it arguable.
- Test one idea per statement.
- Avoid absolute words like “always” and “never,” which learners learn to mark false.
- Keep true and false statements similar in length.
Matching-type items
Matching items work when everything in both columns belongs to one category. A single column mixing dates, people, and definitions gives the answers away.
- Keep each list homogeneous — all events, all terms, all functions.
- Put a few more options than premises so the last answers cannot be found by elimination.
- Keep the whole set on one page.
Completion and short-answer items
These items test recall without cues, which makes them valuable and also prone to multiple correct answers. Plan for that when you write the answer key.
- Put the blank near the end of the sentence.
- Ask for one key word or number, not a phrase that can be worded many ways.
- List every acceptable answer in the key before checking begins.
Essay and problem-solving items
Essay items are where learners can show the most, and where scoring is most at risk of being inconsistent. The fix is to write the scoring guide at the same time as the question.
- Tell learners exactly what the answer must do: “Explain two causes and give one example of each.”
- Give the points and a suggested length.
- Write the rubric before giving the test; our guide on making a rubric for a performance task shows how.
To raise the cognitive level of any item, our Bloom’s taxonomy question stems give ready-made openings for every level.
Step 5: Review and edit the items
The first draft of any item is a draft. A structured review catches the problems you cannot see because you already know the intended answer. Set the items aside for a day if you can, then check each one against this list.
- Does it match its competency and cognitive level in the TOS?
- Is there exactly one defensible correct answer?
- Could a learner who knows the content still get it wrong because of the wording?
- Does it give away the answer to another item?
- Is the vocabulary what you actually used in class?
Ask a colleague to answer a sample of items without the key. If AI drafted any items, review them with extra care — our guide to what AI gets wrong when writing test items lists the specific problems to look for.
Step 6: Assemble the test
Assembly turns a list of items into a test learners can take without confusion. The order, grouping, and layout affect scores more than most teachers expect, especially for younger learners.
- Group items by type into parts, each with its own directions and point value.
- Order items from easier to harder within each part so learners build confidence.
- Number items continuously across parts so the answer sheet and TOS stay simple.
- Keep each item on one page, including its options or diagram.
- Check the time. As a rough guide, allow about 30 seconds per true-or-false item, about a minute per multiple-choice item, a couple of minutes per short-answer item, and ten minutes or more per essay — then adjust for your learners.
Step 7: Prepare the answer key and scoring guide
The answer key is part of the test, not an afterthought. It decides whether two teachers checking the same paper give the same score, and it is the document a learner’s parent will ask to see.
Take the test yourself without looking at your intended answers, then compare. Fix every disagreement before printing. Check that correct letters are spread evenly with no pattern, list all acceptable answers for completion items, and attach the rubric for every essay. For the full set used in a DepEd term exam, see our guide to the periodical test with answer key and TOS.
Step 8: Try it out and prepare alternate forms
A test that has never been taken by anyone is still a hypothesis. Even a small try-out reveals unclear items, timing problems, and answer-key errors while there is still time to fix them.
Our guide to pilot testing an exam explains how to do this with limited time. If sections will take the test on different days, prepare parallel forms built from the same TOS so the versions are genuinely equivalent.
Step 9: Administer and score consistently
Consistent administration is part of reliability. Every section should get the same directions, the same time, and the same conditions, and every paper should be checked against the same key.
A separate answer sheet speeds up checking considerably — see our printable answer sheet templates — and our guide to checking 100 test papers fast covers methods that save hours. Record scores the same day; a paper that waits two weeks to be recorded is a grade at risk.
Step 10: Analyze the results and improve the test
The last step is the one most often skipped, and it is what turns a one-time test into a better one next term. Two quick analyses tell you most of what you need.
The mean percentage score (MPS) shows how the class did overall and per competency. An item analysis shows which items were too easy, too hard, or failed to separate stronger from weaker learners — and those items should be revised or replaced before the test is used again. The Filipino version of the item analysis guide is Paano Gumawa ng Item Analysis.
How to construct a test under DO 15, s. 2026
The principles above apply to any test, but DepEd teachers also work within the structure set by DO 15 for SY 2026–2027. A few points affect how you construct and use summative tests this year.
- Each term has two summative tests and a term examination. ST1 and ST2 cover their own blocks of lessons; the term exam covers the whole term. See the three-term assessment calendar.
- Inside the Examinations component they weigh 30%, 30%, and 40%, each converted to a percentage first, so the number of items never decides the weight.
- Formative assessments are not graded. They guide teaching and do not enter the computation.
- ST1 results start remediation for learners consistently below expectations, as early as Week 5.
Using AI in test construction without losing quality
AI can save real time on the most tedious parts of this process — drafting items, writing distractors, and formatting the paper. It cannot know what you actually taught, how your learners talk, or which items are subtly wrong. Treat it as a fast first-draft writer and keep every decision yours.
Give it your own material rather than a topic name, review every item against the TOS, verify the answer key yourself, and never paste learners’ names or records into an AI tool. Our guide to student data privacy when using AI tools explains what to keep out.
Where Skoolari fits
Skoolari follows this process rather than replacing it. You upload your own module or lesson notes — PDF, Word, PowerPoint, or text — choose the item types and difficulty, and it drafts the test with an answer key for you to review. In DepEd mode, the TOS-Based mode builds the test from your competencies, days, and cognitive-level plan and produces the Table of Specifications with it. In College mode, each item is tagged with the source it was drawn from so you can verify it, and a TOS can be built from the finished paper.
After the test, the class record computes each test’s MPS and result analysis, and the Item Analysis report flags items to revise. You can start free; the plans page lists what each plan includes.
Frequently asked questions
These are the test construction questions teachers ask most often. Each answer points to the step above where the topic is covered in more detail.
How many items should a test have?
Enough to measure each competency reliably in the time available. Very short tests are less reliable, because a single lucky or unlucky item changes the score a lot. Your school may set item counts for summative tests and term exams; within that, give each competency enough items to show what learners can do.
Should every test have a table of specifications?
Every summative test benefits from one, and many schools require it. For a short formative quiz, a quick mental blueprint is usually enough — just make sure the quiz covers the lesson you meant it to check.
How do I make a test harder without making it unfair?
Raise the cognitive level, not the obscurity. Ask learners to apply, analyze, or evaluate what they learned in a new situation, rather than testing trivial details or using confusing wording.
Can I reuse last year’s test?
You can reuse good items, but rebuild the test around what you taught this term, and revise items your item analysis flagged. A test copied whole from last year often includes competencies that were not taught this time.
DepEd assessment rules for SY 2026–2027 are still being clarified by memoranda. Check the DepEd-specific points above against the current issuances and your school’s policies.