Open response vs multiple choice: choose the evidence your quiz needs

Choose open response or multiple choice by the evidence you need: a selected conclusion, a calculation or a learner-generated explanation. A multiple-choice item can ask learners to apply a rule to a novel case. An open prompt can ask them to copy a fact. The learning goal, answer key, time available and follow-up action determine which is useful. A mixed quiz often collects better evidence than a set of questions forced into one format.

The OECD overview of assessment tasks discusses the efficiency of objective items and the richer reasoning that extended responses can reveal. Its work on complex skills explains why more open tasks may be necessary when the goal is to observe a process. These sources do not say every open question is good or every selected answer is weak. They support matching format to inference.

Define the science claim you want to test

Consider a lesson in which students compare two experiments about plant growth. Both show plants grown with different amounts of light, but one also changes the amount of water. The goal is to decide which claim the data support and explain a limitation. First, ask students to identify the variable changed in experiment A. If the diagram and answer are clear, multiple choice or a short exact answer is efficient. Next, ask which conclusion is supported. Carefully written options can distinguish a narrow claim from an unsupported generalisation.

Finally, ask: “Why does experiment B not isolate the effect of light? Name the second variable and explain its role.” This is where a short open response earns its marking time. A student might pick the right conclusion from options without being able to explain confounding. The third answer shows whether they can connect evidence to method. If that reasoning is the true objective, do not let two quick correct clicks overshadow a weak explanation.

The guide to writing good multiple-choice questions helps make the closed item unambiguous. Here the contrast between the two experiments is essential: if the diagram is too small or a label is missing, a wrong answer may be a problem with the task, not the concept.

Use each format for the part it can show

Multiple choice can sample several objectives in a short period and can be marked against a reviewed key. Its options, however, make recognition easier than production. Open response invites a learner to generate a reason, but it also introduces writing demands that may not be part of the science goal. Keep the prompt short, state the expected evidence and consider whether another way to explain is needed under a learner’s support plan.

Do not ask an open question merely to appear rigorous. “What colour is the bar on the chart?” gains nothing from a text box. Do not ask a closed question merely to save marking time when the real target is a nuanced argument. The format should allow the intended mistake to become visible, so that you can address it in the next lesson.

Write a key and a rubric before sharing

For the closed item, write the correct response in words, then inspect every distractor. Is one also arguably correct because of the experiment’s wording? Does a wrong option reflect a real confusion, or is it obviously silly? Have a colleague try the item without hearing your intended answer. If they reasonably defend another option, revise the stem or switch to an open explanation. The auto-grading guide explains why automatic comparison still depends on a correct, fair key.

For the open answer, set two criteria: names water as the additional changed variable and explains why a growth difference cannot be attributed to light alone. A third criterion might ask for a concrete improvement, such as holding water constant. Accept different phrasing that meets those criteria. A response that merely says “the experiment is unfair” needs a prompt to specify what changed and why it matters.

Feedback should show the next action: “You noticed that water changed; explain how that prevents a clean comparison of light conditions.” The Education Endowment Foundation feedback guidance emphasises responding to a specific gap. An automatically assigned total may obscure that the learner knows the variable but cannot yet explain the inference.

Worked plant experiment: build the data before comparing formats

The following numbers are invented for question design, not reported plant-growth results. Assume the same plant variety, initial height of 10 cm, growing period, soil, temperature and measurement method. In experiment A, daily water is held constant; in B, both daily light duration and water change. The small example has one plant per condition, so it also lacks replication.

A closed item can now ask: “Which experiment keeps water constant when comparing light duration?” The key is A. Another can ask for the growth difference in A: 9 − 5 = 4 cm. Neither correct selection alone shows that the learner can explain why B is confounded. Keep that distinction in the feedback: a calculation and an inference are different evidence.

For the open item, ask: “Why can the growth difference in B not be attributed to light duration alone? Use the table, then suggest one change.” Award one point for identifying that water also changes, one for explaining that its effect cannot be separated from light in this comparison, and one for a suitable control such as giving both plants the same daily water. Do not award the causal explanation point merely for naming water.

Keep the conclusion narrower than the question’s appearance

Experiment A is a better controlled comparison of light duration under the stated conditions. Four extra centimetres in these invented plants do not establish an effect for every species, an optimal light schedule or a reliable population estimate. Replication and appropriate experimental controls would be needed for stronger claims. If you ask for this limitation, add it explicitly to the rubric rather than marking students down for an unstated requirement.

Finally, test transfer with a new table where temperature and water, rather than light and water, change together. Ask students to identify the two changes and propose one controlled comparison. That check tests whether they can reuse the confounding argument; repeating “B is unfair” on the original table could simply reproduce the feedback.

Budget the time for reading answers

If you have 30 students and ten open questions, you may not be able to give meaningful feedback on all of them. Choose one open prompt that reveals the most important reasoning, then use closed items for prerequisite checks. A shorter quiz with a well-reviewed explanation can be more useful than a long mixed test whose open responses never get read. Tell students what is being marked automatically and what awaits teacher review.

Accessibility also affects the choice. A long text response can be hard for a learner whose barrier is writing rather than scientific reasoning; a colour-coded diagram can block another learner from both formats. The quiz accessibility checklist shows how to test images, instructions and timing. Give an equivalent response route where required, without quietly changing the learning objective.

For the transfer task above, keep the calculation and reading demands comparable to the first table. Otherwise a new error may reflect a harder calculation or unfamiliar vocabulary as well as a difficulty with confounding.

Build a mixed quiz in Skolina

Skolina lets you combine closed and open questions and share them by link or QR code. Advanced per-question reports require Starter, Pro or eligible institution access. Objective keys can be marked automatically; read the causal explanations yourself. Check the current plans if you need several versions for groups or sessions.

To try it, create a Skolina teacher account, choose a real learning goal and write one closed item for a prerequisite and one open item for the reasoning. Before sharing, record what each answer will let you conclude and what it will not. That record is more useful than choosing the format by habit.