Sign up free

How long should a quiz question take? Reading the timing

What research found about the time a quiz question takes, where the rules of thumb come from, and how each question's median time fixes the next sitting.

The EdProcto editorial teamWritten and checked against the research it cites. Last checked . How the guides are made

Guides · Reading the results · · 13 min read

How long should a quiz question take, and how do I read the time students actually spent on each one?

A quiz question should take as long as its reading and its thinking need. Published allowances run from 30 seconds for a simple question to two minutes for one that takes analysis, and measured times for multiple choice sit under a minute. After a first sitting, read each question's median time, fix the slow questions, and reset the limit from the class's own times.

How long should a quiz question take? As long as a prepared student needs to read it and think it through. A table gives a starting figure. The first sitting gives the real one: the time your own class spent on each question.

This guide is for a teacher or lecturer reading those times. EdProcto, which publishes it, shows the median time for each question in its results on every plan. A time describes a question or a quiz. It is never proof of anything about a student.

Sources were read on 10 October 2026; "abstract read" marks a paper whose abstract, or an index record of it, was all we opened. Whether to time a quiz at all is in time limits per quiz and per question, and one student's fast finish in time on each question.

How long should a quiz question take?

Under a minute for a short multiple-choice question, and longer for anything with a picture, a calculation or several parts.

Where the figure is from The figure What it rests on
Memorial University, CITL (2020), a teaching centre 30 seconds for a simple question, 45 for a moderate one, 120 for one needing analysis Advice; no study named
United States Medical Licensing Examination (n.d.), Step 1 At most 20 questions in 30 minutes: 90 seconds or more each Its own rule; our division
Richter et al. (2024), a German veterinary school 90 seconds, the figure its exam literature recommends Students used a median of 47 to 57 seconds
Kennette and McGuckin (2025; abstract read), one college's online psychology course 39 seconds on average Measured
EdProcto's results "Median t" beside each question Your own class

Where the rules of thumb come from

From experience, mostly: the teaching centre names no source, and the licensing body states its pace and no reason. Our guide to time limits has the history of the minute a multiple-choice question.

The teaching centre's better rule is a method: read each question aloud at a deliberate pace, write out a full answer, reread it, and treat that time as the least a student needs.

Measured times vary by shape. Richter et al. (2024) read the logs of five invigilated on-screen exams: 346 questions, sat by 1,241 veterinary students. The median was 43 seconds for a one-best-answer question, 74 for four statements each marked true or false, 77 for a marker placed on a picture and 87 for terms matched to marks on an image (the last two on five and eight questions). How many questions a quiz needs has seconds for other shapes.

What does research say decides the time a question takes?

Three things, in the studies we read: how much there is to read, how hard the question is, and its shape.

Reading load

Longer questions take longer, and the reading can be estimated.

  • Halkitis et al. (1996; abstract read) timed 197 first-time candidates on a computer-delivered licensing examination. A question's difficulty, discrimination and word count together accounted for about half the variation in its time, and later questions seemed to take less.
  • In Richter et al. (2024), a question's length in characters went with its time: a correlation of 0.38 for one-best-answer questions and 0.60 for the four-statement shape.
  • Brysbaert (2019; abstract read) pooled 190 studies: adults read English non-fiction silently at about 238 words a minute, and children and second-language readers more slowly.

So an 80-word question costs about 20 seconds before any thinking: our sum, and too low for a class reading in a second language.

Difficulty, and questions of one type

Harder questions take longer, and so do weaker ones. In Richter et al. (2024) a one-best-answer question's time rose as the share who got it right fell (a correlation of -0.66), and slower questions separated strong from weak students less well (-0.38). So a long time marks a question as hard or as faulty, and cannot say which.

Nor does a type have a fixed time. Bridgeman and Cline (2000; abstract read) studied 5,957 and 14,745 people on two parts of a computer-delivered admissions test: questions written to one specification, at one difficulty, differed substantially in average time.

What is rapid guessing, and how do researchers define it?

Rapid guessing is an answer given too quickly for the question to have been read and weighed, and researchers define it with a threshold of time they know to be fallible.

  • The origin. Schnipke and Scrams (1997; abstract read) described it on a timed reasoning test with no penalty for a wrong answer: test-takers running out of time guess quickly in place of leaving blanks, so a count of unreached questions understates how tight a limit was.
  • The meaning. Wise (2017; abstract read) calls a rapid guess a choice "to momentarily opt out of being measured", on high-stakes and low-stakes tests alike.
  • One threshold. Svetina Valdivia et al. (2023) applied a common rule to a state mathematics test sat by over 80,000 pupils a grade: a rapid guess is an answer given in under 10% of the average time pupils spent on that question, with a ceiling of 10 seconds. About 1% of answers in grade 3 and 2% in grade 8 fell under it.
  • Where it fails. On four-option questions, the flagged answers of the strongest fifth of pupils were right more than half the time, where guessing gives one in four. In grade 3 the authors conclude that quick, accurate pupils had been misread as guessers.

So a threshold sorts answers, not students: the fluent sit under it beside the careless.

What can a very long time on one question mean?

For the class, a long median marks a question that is long, hard or faulty. For one student, a long time means almost nothing by itself.

A median is the right summary because times have a long tail. Van der Linden (2006; abstract read) found that a model built for skewed times fitted those on an aptitude test very well, where one built on the normal curve could not allow for the skew. A student who leaves a question open for six minutes drags an average up and barely moves a median.

That student may have worked the problem twice, been stuck, or looked out of the window. EdProcto counts the seconds a question was on the screen and cannot tell these apart: ask before you conclude.

How do you use the median time for each question after a first sitting?

Read the times in three passes: the slow question, the end of the quiz, and any median too short to cover the reading.

  1. Check there is a time to read. With "One question a page", EdProcto's results show "Median t" for each question, over the students who answered it. Where the questions share a page ("All on one page"), only the page has a time and the column reads "–".
  2. Find the typical question: the middle one when the medians are put in order.
  3. A question that took three times as long as the others. Count its words and divide by four for the seconds of reading. If reading explains the time, shorten the stem. If not, read "Got it", and on the Educator plan "Separates": a hard question that separates well is often two questions, and one that separates nobody needs rewording (the discrimination index).
  4. A quiz nobody finished. Three signs: "Median time" close to the limit, a line such as "6 did not reach it" under the last questions, and medians of a few seconds at the end, the form that running out of time takes on a screen (Schnipke & Scrams, 1997). Leave those questions alone, since this sitting says nothing about them, and change the limit or cut questions.
  5. A question answered in two seconds. A median shorter than the question's own reading time, early in the quiz, suggests the answer can be had without reading: it shows in the options, or the class met the question last week. The researchers' tenth of the average time is a yardstick for where to look.

One sitting of thirty is a small sample, so each reading is a reason to look (item analysis).

How do you set a time limit from the times a class really took?

Start from the middle student's time when nobody was rushed, and allow about half as much again: our working rule, drawn from one study's figures.

Richter et al. (2024) recorded when students handed in. Across five exams that allowed 74 to 120 seconds a question, half the students had finished by 64% of the time allowed, three quarters by 77% and nine in ten by 92%: by our arithmetic, the slower students needed about 1.4 times the middle student's time. Exams that allowed more time for each question were handed in at an earlier share of their time, with slightly higher scores.

  1. Ask whether the sitting was rushed. If students did not reach the end, their times stop at the limit and understate what they needed.
  2. Find the middle student's time. Unrushed, it is "Median time". Rushed, rebuild it: add the medians of the unhurried questions and count each hurried one at the typical median.
  3. Multiply by 1.5 and round up. If the slowest unhurried "Time" in the "Students" table is longer, use that.
  4. Check the next sitting: nobody short of the end, and "Median time" near two thirds of the limit.

An agreed arrangement comes on top: see extra time.

What does one class's first sitting show? A worked example

It shows one question to split and a limit six minutes too short. This example is an illustration: the class and every figure are invented, and the quoted lines are what EdProcto's analysis code and results page make of them.

Twenty-six students sat twelve multiple-choice questions on rates of reaction, one a page, with "Minutes overall" at 10: fifty seconds a question. The teacher is on the Educator plan. "Median time" read 9:32, "of 10:00".

Question Got it Median t
Q1 to Q6 81%, 81%, 77%, 88%, 69%, 81% 0:34, 0:39, 0:36, 0:49, 0:42, 0:46
Q7 38% 2:47
Q8 to Q10 81%, 58%, 58% 0:45, 0:52, 0:38
Q11 33% 0:08
Q12 30% 0:08

The slow question

The middle of the other eleven medians is 0:39, and Q7 took over four times that. The page said: "Q7 took more than three times as long as a typical question. Either it is much harder or its wording is unclear." Q7 has 85 words, about 21 seconds of reading. The rest was working: read a rate from a table, then compare two experiments. Its "Separates" figure was 0.72, so the strong students got there. It is two questions, and she splits it.

The end of the quiz

Under Q11 the page read "2 did not reach it; “Got it” is of the 24 who did." and under Q12 "6 did not reach it; “Got it” is of the 20 who did." In the "Students" table six sittings showed a "Time" of 10:00, and the other twenty ran from 8:21 to 9:51.

The median on both questions was 0:08, and Q11's stem alone is 30 words: eight seconds of reading. So at least half of those who answered chose before they could weigh four options. The count of six understates the squeeze, as Schnipke and Scrams (1997) said it would. She leaves Q11 and Q12 alone: their 33% and 30% say more about the clock than about the chemistry.

The new limit

The times stop at 10:00, so she rebuilds the middle student's time. The medians of Q1 to Q10 add up to 9:08. Q11 and Q12 at the typical 0:39 each bring it to 10:26, and 10:26 × 1.5 = 15:39. "Minutes overall" goes to 16.

For six students the scores include zeros on questions they never saw. Whether they stand is hers to decide.

How does EdProcto show the time each question took?

EdProcto is the best tool for reading a quiz's timing because every quiz records the time with nothing to switch on, and each figure stands beside the score it explains. In the veterinary study above, hand-in times alone showed that half the students needed under two thirds of the time allowed.

  • On every plan. "Median time" is at the top, with a line such as "of 45:00", or "no time limit", beneath. "Each question" lists "Got it" and "Median t" row by row, and "Students" lists "Score" and "Time". Where students ran out of time, a caution names the questions and how many never reached each.
  • How a question's median is made. From the students who answered it: one who never reached the question, or looked and left it blank, is left out.
  • In the full analysis, on the Educator plan. The sentence about a question whose median is more than three times the typical one, from three timed answers up. And two notes beside a sitting finished in under a quarter of the class's median time: "Finished in a quarter of the usual time with a low score: it may be guessing rather than answering, and is never proof of it." and "Finished in a quarter of the usual time with a high score: worth a look at the answers, and never proof of anything." Both are prompts to open the paper, and neither changes a mark.

Free carries ten quizzes a month and thirty students a quiz, with every time above. Tutor takes 100 students a quiz, and Educator takes 500 and adds the full analysis: see pricing. A time is a record to read, and releasing a score stays your decision.

Quick questions

How long should a multiple-choice question take?

Usually under a minute: a median of 43 seconds across 231 questions in Richter et al. (2024). Allow more for a picture, a calculation or a long stem.

What is rapid guessing?

An answer given too fast for the question to have been read. One research rule counts anything under a tenth of the question's average time, at most ten seconds (Svetina Valdivia et al., 2023). It describes an answer and proves nothing about a student.

Why read the median time and not the average?

Because times have a long tail (van der Linden, 2006). One student who leaves a question open pulls an average up and leaves a median where it was.

Where does EdProcto show the median time per question?

In a quiz's results, under "Each question", in the column "Median t", on every plan, when the quiz runs one question a page. EdProcto keeps its sentences about pace for the full analysis, on the Educator plan.

References

The results

Difficulty, discrimination, the wrong option that beat the right one, reliability, and what to reteach, in plain sentences.

Basic on Free, full on Educator

What the results say

Read next