A question bank for schools: from good questions to marks in the gradebook
Why departments keep a shared question bank, how to tag a question so a colleague can find it, how to write wrong answers and mark schemes that hold up, how to put together a fair test, run a quick quiz in the lesson — and get the marks into the gradebook without typing them twice.
A question bank for schools is a department's shared store of questions, each tagged with what it checks, so any teacher can build a test or a quiz from it. In short: tag every question with subject and year, learning outcome, level of demand, type, time and points; write each wrong answer from a real mistake; give every open question a mark scheme before anyone answers it; have a colleague check a question before it is shared; build each test from a plan of outcomes and levels; use the same questions for low-stakes quizzes; let software mark the closed questions and suggest the marks, but send nothing to the gradebook that nobody has checked.
The advice works with any tool. The "In Bildena" paragraphs describe the Learning module of Bildena School as it is today, limits included.
Why departments build a question bank
Three Year 8 classes with three teachers and one scheme of work need comparable tests; a bank makes that the normal case. A good question with good wrong answers and a mark scheme takes time to write, and written and checked once, it serves for years. After a few classes you also know how many get it right and which wrong answer they choose. The honest cost: a bank pays off when several teachers teach the same subject and year, and only if the questions are tagged. A folder of old papers is an archive, not a bank — nobody can search it for "percentages, apply, two minutes".
How to tag questions in a question bank
| Tag | What it holds | Example |
|---|---|---|
| Subject and year | Where it belongs | Maths · Year 8 |
| Learning outcome | A line of the curriculum, with its code | N3 — increase and decrease by a percentage |
| Topic | An agreed word where there is no curriculum line | Percentages |
| Level of demand | I reproduce · II apply · III reason | II |
| Type | How it is answered | Single choice |
| Time and points | Minutes a typical student needs; points | 2 min · 2 points |
| Status | Draft · in review · shared · retired | Shared |
Tag while writing, from a fixed list: "%", "percent" and "percentages" are three topics to any search. Give each question one main outcome, and retire questions rather than delete them — old tests used them.
In Bildena the bank is Learning › Question bank. Each question is tied to lines of the year's curriculum — the lines Lesson plans teach — with a level (I Reproduce · II Apply · III Reason), an estimated time and its points; where the year has no curriculum in Bildena, a topic stands in. The list filters by subject and year, by outcome, and by whether an outcome is taught now, was taught this year or not yet, and it counts how many of the outcomes taught so far have at least one question.
Writing good questions: multiple choice, short answer, extended answer
Multiple choice: the wrong answers do the work
Each wrong answer (distractor) should be the answer a student gives after a typical mistake:
A jacket costs €80. In the sale the price is reduced by 25%. What is the sale price?
| Choice | The mistake behind it | |
|---|---|---|
| €60 | right | — |
| €55 | wrong | takes off 25 euros instead of 25 percent |
| €20 | wrong | gives the discount, not the new price |
| €100 | wrong | adds 25% instead of taking it off |
Weak alternatives would be €1,000 (nobody picks it), "none of the above" (ruling out the rest shows nothing about percentages) and €59.99 (a trick about rounding, not the outcome). When a third of the class picks €55, you know what to re-teach. The item-writing guides agree: put the whole problem in the stem; take wrong answers from errors in old scripts; three good options are usually enough (a meta-analysis of 80 years of research, Rodriguez, 2005); avoid "all of the above" and use "none of the above" rarely (Haladyna, Downing and Rodriguez, 2002).
Short answer: decide in advance what counts
List the accepted answers before the test: spelling variants, units ("€60", "60 euros"), synonyms. Decide what counts as nearly right — a spelling slip is usually fine in science, not in a spelling test. For a number, give the tolerance and the unit. Anything not on the list goes to the teacher, not straight to zero.
Extended answer: the mark scheme comes first
A points-based mark scheme lists what earns each point:
Explain why an ice cube melts faster on a metal tray than on a wooden board. (4 points)
- Both surfaces are at room temperature, warmer than the ice. (1)
- Heat flows from the warmer surface into the ice. (1)
- Metal is a good conductor, so heat reaches the ice quickly. (1)
- Wood is a poor conductor, so less heat flows each second. (1)
Level descriptors suit writing judged as a whole. For "Should our town close the high street to cars? (6 points)": level III (5–6) weighs both sides with evidence and reaches a justified conclusion; level II (3–4) explains arguments with some evidence; level I (1–2) states points without explaining them. Write the scheme before the question is used, and answer the question yourself against it.
In Bildena there are thirteen question types. Bildena marks eight closed ones itself the moment a student hands in: single choice, multiple choice, true/false, number (with a tolerance and a unit), fill the gaps, matching, order and points on a grid. A short answer is marked by Bildena when you give accepted answers, and one that is nearly right waits for your yes or no. Worked answers, extended answers and answers on paper are the teacher's, with a marking scheme of steps or level descriptors. For each wrong choice the wizard asks "why a student picks it", and its last step shows what is missing and whether a similar question is already in the bank. A new question is a draft only its writer sees; sent to the department, a second teacher of the subject checks it before everyone uses it. Once students have answered a question, a change becomes a new version, and tests already written keep the one their students saw.
How to create an exam paper from the bank
Start from a plan (a blueprint) that gives each outcome its points at each level — here for a 45-minute Year 8 test on percentages, 30 points:
| Outcome | I | II | III |
|---|---|---|---|
| Find a percentage of an amount | 4 | 3 | — |
| Increase and decrease by a percentage | 3 | 5 | 2 |
| Reverse percentages | — | 4 | 3 |
| Percentage change in context | — | 3 | 3 |
| Total | 7 | 15 | 8 |
- Take the outcomes taught since the last test, not the chapter headings.
- Fill each cell from the bank; prefer questions the class has not met in a quiz.
- Add up the minutes and leave a margin: a paper whose questions fill the whole lesson is too long.
- Mix closed questions, which cover a lot quickly, with one or two open ones that show reasoning.
- Make a second version for neighbours: the same questions in another order, or variants with other numbers.
- Prepare arrangements (extra time, large print, reading aloud), the mark scheme and the grade boundaries before the day.
- Ask a colleague to take the test: every ambiguity found now saves an argument with 28 students later.
In Bildena Learning › Exams › Add exam starts with the type — the written kinds of your school system, such as a class test or short test, or a diagnostic test, oral exam or quiz — which sets how it counts. The plan step lists the outcomes taught since the last written exam, read from Lesson plans; you give each its points at levels I, II and III, and every cell shows how many questions the bank has and how many this class has not seen. The questions step fills the exam from the plan, or you pick from your own and colleagues' shared questions; a random block gives each student different questions of one topic and level. On paper, Bildena prints a paper per student with the name filled in, versions A and B, large print where an arrangement asks for it, and a teacher's copy with answers, marking scheme and grade key. In the app, students answer on their own or on school devices while you supervise; extra time from an arrangement is applied by itself, and exam mode tells you when a student leaves the exam.
Where tests sit in the school's rules
England has no national rule on how many tests a week; since levels were removed in 2015 the DfE has left assessment to schools' own policies. In Germany each state sets its own rules, and they differ:
| State (text) | Per day | Per week | Notice | Returned |
|---|---|---|---|---|
| Bavaria, Gymnasium (GSO §§ 22(4), 25(1)) | one major test | should not exceed two | at least a week | within two weeks (three in some years) |
| North Rhine-Westphalia, primary and lower secondary (BASS 12-63 No. 3.2; APO-S I, VV 6.1.2) | one, and no other written test | two | "in good time" | within three weeks; only then the next one |
| Baden-Württemberg, secondary (NVO § 8) | should not exceed one | should not exceed three | as a rule | before the next test in the subject |
| Hesse (VOGSV §§ 28(2), 33) | one | three (two in primary) | five school days | as a rule within three school weeks |
Two details are easy to miss: short tests usually fall under other rules than class tests, and limits apply to each student — someone in a language or religion group can be over the limit in a week when the class as a whole is not.
In Bildena a state's rules are data. For Bavaria (Gymnasium, Realschule, Mittelschule) and North Rhine-Westphalia's lower secondary they were read from the legal texts, and a school can only make them stricter; elsewhere, in Germany and outside it, the school sets its own numbers. Bildena counts each student's tests on the day you choose. A day over the limit needs a reason and school management's OK; a day the class is away with the school cannot be chosen, nor, where the state requires it, a new class test before the last one is given back. Once announced, the test appears in the students' and families' This week and calendar, with its topics if you wish; an unannounced test is never shown before it is written.
Running a quick quiz in the lesson
A short quiz shows what a class remembers, and recalling helps memory in its own right: in a well-known experiment, students tested on a text remembered more of it on the later tests than students who reread it, although rereading won on a test five minutes later (Roediger and Karpicke, 2006). For a classroom quiz that helps:
- Low stakes: no mark, no names on the board, no leaderboard — a ranking rewards speed and shows the same students at the bottom every time.
- Short: four to six closed questions, about half a minute each.
- Pause before revealing: show how the class answered and talk about the most common wrong answer.
- From the bank: the quiz checks the outcomes the test will — but a question met in a quiz tells you less in the test itself.
In Bildena a quiz is a practice exam: a class quiz runs live in the lesson, online practice stays open at home until a day you set. Live, the board shows one question, the clock and how many have answered — never who. Students answer on their devices, where they see only the letters while the choices are on the board, so nobody copies from a neighbour's screen; you reveal how the class answered and move on. A question stays open between 10 seconds and 10 minutes (30 seconds unless you change it). There is no mark and no ranking; each student sees their own answers. The results show the share right, the hardest question and each question's most common wrong answer; each student's score is for the teacher only, and when a question went badly the results offer to set practice homework.
Marking and feedback
Let software mark the closed questions. Mark the open ones question by question rather than paper by paper, so the twentieth answer is judged like the first, and without names where you can; the same mistake deserves the same comment. And look at the questions as well as the students: if those who did well overall and those who did badly get a question right about equally often, it tells you little — rewrite or retire it.
In Bildena open answers are marked one at a time, question by question, without names unless the exam says otherwise, in an order that does not follow the alphabet. Ticking the steps of the marking scheme adds up the points, and saved comments can be used again. The results give each student's points per part and, per question, the share of points and the most common wrong answer. "Practice for this question" opens an online practice of its topic — the question and up to four more that Bildena marks — for the students who got less than half its points. In the bank, each question shows what students picked, with the reason written for each wrong choice, and a "tells little" flag when stronger and weaker students get it right about as often.
Getting the marks into the gradebook without retyping
Points on paper, a total, the grade boundaries, a mark, the gradebook, the report: each copy is a chance for a typo. Enter points per task once, let the boundaries turn them into marks, check the cases near a boundary, and send. Absent students get a make-up, not an empty cell that later reads as a zero.
In Bildena the points come from marking in the app, or for a paper test you enter each student's points per task. Bildena suggests each mark from the points and the grade key; you check it, change any and send the marks to the gradebook, where the exam counts with its type's weight. Students and families see nothing until you send. Absent students are set for the make-up. Where the school's rule asks for it (the default in a few German states), the marks go to school management first when too many papers are below a pass. Sending again changes the marks, and the change is recorded.
What Bildena does not do
- It does not import question files (Word documents, Moodle or other quiz formats); questions are written in the wizard.
- It does not write questions for you with AI.
- The bank is the school's own: no sharing with other schools, no bought-in question library.
- Quiz results are practice and never go into the gradebook.
- Exam mode does not lock the device: it opens full screen and tells you when a student leaves.
- Learning is part of Bildena School, not of Bildena Scheduler (the timetable on its own).
Tests are one part of what a week asks of students; how the timetable spreads the other part over the evenings is the subject of Homework load: the part of the timetable students take home.
Frequently asked questions
What should a question bank for a school contain?
Questions a colleague can find and use without asking the writer: tagged with subject, year, outcome, level, type, time and points, with the right answer, the reason behind each wrong choice and a mark scheme for open questions.
How many options should a multiple-choice question have?
Three plausible options are usually enough. A fourth is only worth writing if students really make that mistake.
Can quiz results count towards a grade?
They can, but a quiz that counts is no longer low-stakes, and students start hiding mistakes instead of showing them. In Bildena a quiz is practice: no mark, no ranking, nothing in the gradebook.
Learning — lesson plans, homework, exams and the question bank — is one of the modules of Bildena School. In the free month you try every module, first with the sample school, then with your own data, no card needed.
