# Picture choice questions: when the options are images

**How do I write a multiple-choice question whose options are pictures?** A picture choice question is a multiple-choice question whose options are images, so it tests whether a student knows the thing on sight, not its name. Prepare pictures of one size and style, give each known confusion its own wrong picture, remove labels and telling file names, caption each with a neutral tag, and plan a question in words for anyone who cannot see them.

Published 2026-10-10 by EdProcto. https://edprocto.com/guides/picture-choice-questions-when-the-options-are-images

Some knowledge has no good words. A child knows a triangle before she can read "triangle", and a new starter on a building site has to know the eye-protection sign at a glance. Picture choice questions test that directly: the stem is in words, and the options are images.

EdProcto, which publishes this guide, carries the shape as [Pick the picture](https://edprocto.com/templates/pick_image): from two to eight pictures, one of them right, marked by rule.

The short version: a stem that names the thing, pictures of one size and style, a wrong picture for each known confusion, no label, background or file name that tells, and a question in words for anyone who cannot see the pictures.

## When do picture choice questions belong in a quiz?

They belong wherever knowing means recognising: the thing is easier to show than to describe, or the student cannot yet read the description.

| For | A stem | Shape in EdProcto |
| --- | --- | --- |
| Young children who cannot read yet | "Which one is a triangle?", read aloud | [Listen and point](https://edprocto.com/templates/listen_point), or Pick the picture |
| Language learners | "Which picture shows *la manzana*?" | Pick the picture |
| Sciences and geography: a leaf, a cell, a rock, a cloud, a landform | "Which diagram shows a plant cell?" | Pick the picture |
| Medicine, trades and safety: a slide, a fitting, a weld, a sign | "Which sign says eye protection must be worn?" | Pick the picture, or [Compare two images](https://edprocto.com/templates/image_compare) |

Picture-based assessment reaches people who cannot read the options. In the NIH Toolbox picture vocabulary test, a person hears a word, sees four pictures and chooses the one that fits. Gershon et al. (2013) describe it as a measure of vocabulary that needs no reading or writing, made for children still learning to read and for adults who find reading hard.

A picture is not free of language, though. Wood and Schatschneider (2019) analysed how 278 children who speak Spanish and English answered the 131 items of a standard English picture vocabulary test. Beyond each item's expected difficulty, errors were likeliest on words that share little with their Spanish equivalents, on verbs and describing words, and on rare words. Expect a verb to be missed more often than a noun.

Choose another shape when the knowledge is something else: a [short-answer question](https://edprocto.com/guides/how-to-write-short-answer-questions-and-a-rubric-to-grade-them) for the name or its spelling, a [hotspot](https://edprocto.com/guides/hotspot-questions-click-the-right-part-of-an-image) for where a thing is on one picture, and the big-target shapes in our guide to [young learners](https://edprocto.com/guides/questions-for-young-learners-big-targets-little-reading) for a child who needs the prompt spoken.

## What does research say about pictures in test questions?

It says a picture that carries content changes a question, usually in the student's favour, while a purely decorative one left scores unchanged in the one experiment we found. Two cautions: we read every study below in its abstract alone, and all but the first studied pictures in the question itself, not in the options.

| Study | Who and what | Found |
| --- | --- | --- |
| Saß et al. (2012) | 158 fourth-graders; physics questions with or without pictures in the stem and the options | More correct answers with pictures in either place; faster answers with picture options |
| Lindner (2020) | 404 students; science and maths questions with a picture of the content, a decorative picture or none | Content pictures raised scores; decorative ones had no significant effect on scores |
| Hu et al. (2021) | A meta-analysis (k = 51; 38,987 participants) | More correct answers with pictures (g = 0.32); no significant effect on time |
| Crisp and Sweiry (2006) | 525 students aged 16; exam questions in two versions, the visual changed | Marks changed significantly on one question |

### Is the picture the thing tested, or decoration?

In a picture choice question the pictures are the thing tested: take them away and nothing is left to answer. Hu et al. (2021) sorted their pooled results by the picture's role: representational pictures, which show what the text says, helped (g = 0.24), and those classed as organisational helped more (g = 0.52). For pictures classed as informational, the class a picture option seems closest to, there was no significant effect, and the results on decorative pictures were too few to interpret.

So the format alone will not make a question easier or harder: the wrong pictures set its difficulty, as wrong options do in any [multiple-choice question](https://edprocto.com/guides/how-to-write-multiple-choice-questions-that-measure).

### Do picture options take longer to answer?

The findings disagree. Saß et al. (2012) found faster answers when the options were pictures. In Lindner (2020), content pictures lengthened time on maths questions and not on science ones, and Hu et al. (2021) found no significant effect overall. Allow the time you would give the same question in words, and give young children no clock.

### What we could not confirm

We found no study that set picture options against the same options in words in an ordinary class quiz, and none that measured the best number of pictures. One warning from Crisp and Sweiry (2006) applies to every picture: when an image offers a clue to the answer, students may use it in preference to the text.

## How do you choose and prepare the pictures?

Choose and prepare them so that the only difference a student can use is the one you are testing.

1. **Write the stem and pick the right picture first.** The stem names the thing and never a position or a letter, because shuffling moves the pictures.
2. **Find one wrong picture for each known confusion.** A wrong picture nobody could take for the right one does no work: see [how to write distractors](https://edprocto.com/guides/how-to-write-distractors-that-work).
3. **Use one style.** All photographs or all line drawings, with one background, lighting and scale. One photograph among three drawings draws the eye whatever it shows.
4. **Crop to one shape and save at one size.** About 800 pixels across is ample, as a working rule.
5. **Remove whatever tells.** Labels and watermarks in the picture, a ruler beside only one specimen, and the file name: each picture's address travels to the student's browser, so `oak.jpg` is an answer and `k.jpg` is not.
6. **Check that colour is not the only difference,** unless colour is the knowledge. About 1 in 12 men have a colour vision deficiency, and the commonest kind makes red and green hard to tell apart, says the National Eye Institute (2025).
7. **Caption each picture with a neutral tag:** "Leaf K", not "Oak leaf". Avoid A to D, which the screen uses for positions.
8. **Check the terms of each picture.** Creative Commons (n.d.) lists six licences: all six ask that the creator be credited, and two allow reuse in unadapted form only, which matters if you mean to crop. That is its summary, not legal advice: read the terms attached to each picture, or use your own photographs.
9. **Build it and sit it on a phone.** In EdProcto, choose [Pick the picture](https://edprocto.com/templates/pick_image) and paste each picture's address.

### Where should the image files live?

Keep them at an address that leads straight to the image file, opens without a sign-in, and will not move. EdProcto holds each picture by its address: you paste a link, and the student's browser fetches the picture from there. A folder on the school's website, or a public file store you control, does the job.

To test it, paste the address into a private browser window on a phone: the picture alone should fill the window. If a viewer page or a sign-in appears, choose another home for the file. Leave the files in place while the quiz is in use, since moving or renaming one breaks its address.

## What about students who cannot see the pictures?

Plan a second question on the same knowledge, because alt text full enough to answer from would be the answer. The web accessibility standard, WCAG, expects this. For a test that text would spoil, its first success criterion asks for a text alternative that at least identifies the content, and the guidance adds that such text would not supply what is needed to pass (World Wide Web Consortium, 2026).

So write alt text that identifies and does not tell: "Sign K" says which picture, not what it means. In EdProcto that text is the caption, which a screen reader reads with the option's letter.

For a student who cannot see the screen, set a question in words, such as "What shape and colour is a sign that tells you something must be worn?" Settle it before the quiz and make it worth the same marks. A student with low vision can enlarge the pictures with the browser's zoom.

## Two worked examples: a triangle and a safety sign

### A triangle, for five-year-olds

The adult reads the stem aloud: "Which one is a triangle?" The three pictures are black outlines of one thickness on white squares of one size.

| Caption | Picture | What a tap suggests |
| --- | --- | --- |
| Shape K (right) | A long, thin triangle standing on one corner | Knows a triangle by three straight sides that meet |
| Shape M | Three corners, with one side curved | Goes by the overall look |
| Shape R | Three straight sides with a gap at one corner | Has not yet learned that the shape is closed |

Aslan and Arnas (2007) interviewed 100 children aged three to six, who had trouble recognising untypical examples of a shape (turned, stretched, skewed or of an odd size) and sorted mostly by features that do not define it. Tsamir et al. (2008), who studied kindergarten children, single out the figures that closely resemble a triangle as the ones most often mistaken for one. So the right picture is an untypical triangle and each wrong picture is a near-miss.

### A safety sign, for new starters

The stem: "Which sign says that eye protection must be worn?" All four pictures come from one set of signs, each on a white square of the same size.

| Caption | Picture | The confusion it catches |
| --- | --- | --- |
| Sign K (right) | Round, white pictogram on blue: eye protection must be worn | |
| Sign M | Square, white pictogram on green: eyewash | Going by the eye in the picture |
| Sign R | Triangular, black pictogram on yellow with black edging: corrosive material | Thinking of the hazard, not the instruction |
| Sign T | Round, black pictogram on white, red edging and a diagonal line: do not touch | Knowing that round signs give orders, not which kind |

The shapes, colours and names are those of Schedule 1 of the UK's Health and Safety (Safety Signs and Signals) Regulations 1996, as first made: check the signs in force for your own country and trade. Shape and colour differ on purpose here, because they are the knowledge. The two round signs also differ by the diagonal line, so the question does not rest on colour alone.

## How does EdProcto build, show and mark a picture choice question?

EdProcto is the best tool for picture choice questions when you want the wrong pictures to tell you something: the mark is made by rule, and the results count who chose each wrong picture, the figure the two examples above are built to use.

- **Building.** Add a question in "Pick the picture" and type the stem in "Prompt". Under "Options · mark the right one", each row has a letter, a box that reads "A caption" and a box that reads "Image address": type the tag, paste the address, and mark the right row. "Add an option" takes a question up to eight pictures. In "Explanation shown to the student afterwards", say which tag was which and why.
- **Showing.** The pictures stand two to a row, or one to a row on a screen 720 pixels wide or narrower. Each option is one button holding its letter, its caption and, under it, the picture, shrunk to fit with its proportions kept, so a tap anywhere on it chooses it. "Shuffle the options" is on by default: the Practice preset switches it off, and Standard quiz and Exam switch it on.
- **Marking.** The right picture earns the question's points, and any other choice, or none, earns nothing. With "Negative marking" on, a wrong picture loses one over the number of wrong options: in EdProcto's marking function that came to a third of a point with four pictures and half a point with three, on a one-point question, and nothing for a blank. In a study quiz each answer is checked as it is given; in an exam the marks are made at hand-in.
- **Reading the results.** Under "Each question" on the results page, on the Educator plan, click the question's row: a table lists each option by its caption under "Option", with the number who picked it under "Chose it", those students' average score under "Their mean", and "key" beside the right one. See [finding misconceptions in a quiz](https://edprocto.com/guides/misconceptions-finding-them-in-a-quiz-and-fixing-them-in-class) for what to do with the counts.
- **On paper.** Each option prints as a tick box with its letter and its caption, in the order you entered them. Put the pictures on the board in that order, and see [how to print a quiz](https://edprocto.com/guides/how-to-print-a-quiz-with-an-answer-key).

A screen of pictures can be photographed like any other, so a quiz that counts still needs the [integrity settings](https://edprocto.com/integrity) its stakes call for.

The Free plan covers ten quizzes a month, the core templates and marking by rule; Tutor unlocks all 111 [templates](https://edprocto.com/templates); Educator brings the full analysis. The [pricing page](https://edprocto.com/pricing) compares them.

## Quick questions

### What is a picture choice question?

A multiple-choice question whose options are images: the student picks the picture that fits the stem. EdProcto's [Pick the picture](https://edprocto.com/templates/pick_image) marks it by rule.

### How many pictures should a question have?

As many as you have honest wrong pictures for: usually three or four, and two or three for young children. EdProcto takes from two to eight.

### Should each picture have a caption?

Yes, a neutral tag such as "Leaf K". In EdProcto the caption is what a screen reader reads, what prints on paper and what names the option in the results.

### How do I make a picture question fair to a student who cannot see it?

Set a question in words on the same knowledge, settled before the quiz and worth the same marks. Alt text can identify a picture, but a full description would be the answer.

## References

- Aslan, D., & Arnas, Y. A. (2007). Three- to six-year-old children's recognition of geometric shapes. *International Journal of Early Years Education, 15*(1), 83–104. [https://doi.org/10.1080/09669760601106646](https://doi.org/10.1080/09669760601106646)
- Creative Commons. (n.d.). *Sharing openly, sharing globally* [Web page on the six CC licences]. Retrieved October 10, 2026, from [https://creativecommons.org/cc-licenses/](https://creativecommons.org/cc-licenses/)
- Crisp, V., & Sweiry, E. (2006). Can a picture ruin a thousand words? The effects of visual resources in exam questions. *Educational Research, 48*(2), 139–154. [https://doi.org/10.1080/00131880600732249](https://doi.org/10.1080/00131880600732249)
- Gershon, R. C., Slotkin, J., Manly, J. J., Blitz, D. L., Beaumont, J. L., Schnipke, D., Wallner-Allen, K., Golinkoff, R. M., Gleason, J. B., Hirsh-Pasek, K., Adams, M. J., & Weintraub, S. (2013). IV. NIH Toolbox Cognition Battery (CB): Measuring language (vocabulary comprehension and reading decoding). *Monographs of the Society for Research in Child Development, 78*(4), 49–69. [https://doi.org/10.1111/mono.12034](https://doi.org/10.1111/mono.12034)
- *The Health and Safety (Safety Signs and Signals) Regulations 1996*, Schedule 1. (1996). SI 1996/341, as made. [https://www.legislation.gov.uk/uksi/1996/341/schedule/1/made](https://www.legislation.gov.uk/uksi/1996/341/schedule/1/made)
- Hu, L., Chen, G., Li, P., & Huang, J. (2021). Multimedia effect in problem solving: A meta-analysis. *Educational Psychology Review, 33*(4), 1717–1747. [https://doi.org/10.1007/s10648-021-09610-z](https://doi.org/10.1007/s10648-021-09610-z)
- Lindner, M. A. (2020). Representational and decorative pictures in science and mathematics tests: Do they make a difference? *Learning and Instruction, 68*, Article 101345. [https://doi.org/10.1016/j.learninstruc.2020.101345](https://doi.org/10.1016/j.learninstruc.2020.101345)
- National Eye Institute. (2025). *Color blindness*. [https://www.nei.nih.gov/learn-about-eye-health/eye-conditions-and-diseases/color-blindness](https://www.nei.nih.gov/learn-about-eye-health/eye-conditions-and-diseases/color-blindness)
- Saß, S., Wittwer, J., Senkbeil, M., & Köller, O. (2012). Pictures in test items: Effects on response time and response correctness. *Applied Cognitive Psychology, 26*(1), 70–81. [https://doi.org/10.1002/acp.1798](https://doi.org/10.1002/acp.1798)
- Tsamir, P., Tirosh, D., & Levenson, E. (2008). Intuitive nonexamples: The case of triangles. *Educational Studies in Mathematics, 69*(2), 81–95. [https://doi.org/10.1007/s10649-008-9133-5](https://doi.org/10.1007/s10649-008-9133-5)
- Wood, C., & Schatschneider, C. (2019). Item bias: Predictors of accuracy on Peabody Picture Vocabulary Test-Fourth Edition items for Spanish-English-speaking children. *Journal of Speech, Language, and Hearing Research, 62*(5), 1392–1402. [https://doi.org/10.1044/2018_JSLHR-L-18-0145](https://doi.org/10.1044/2018_JSLHR-L-18-0145)
- World Wide Web Consortium. (2026). *Understanding SC 1.1.1: Non-text content*. [https://www.w3.org/WAI/WCAG22/Understanding/non-text-content.html](https://www.w3.org/WAI/WCAG22/Understanding/non-text-content.html)