Formative Assessment: From Evidence to the Next Move

Overview

Formative assessment is an ongoing process in which teachers and students use evidence of learning to improve teaching and learning while there is still time to act. What makes an activity formative is not its label or format, but what happens next: the evidence leads to feedback, student action, or an instructional adjustment.

A quiz, poll, homework task, discussion, or worksheet can all be used formatively. The key question is whether the response helps reveal where students are relative to a learning goal and informs a useful next step. The Iowa Department of Education defines formative assessment as part of instruction, with feedback used to adjust ongoing teaching and learning.

This process is often called assessment for learning. It differs from assessment used primarily to record what students know at the end of a unit. It also puts more value on student thinking than final-answer marking alone. A correct answer may hide uncertain reasoning, while an incorrect answer may reveal a small procedural slip or a productive partial understanding.

Formative assessment is generally low stakes, meaning it carries little or no point value, according to Carnegie Mellon University’s Eberly Center. It does not have to be completely ungraded to be formative. Its defining feature is that the resulting information is used to improve subsequent teaching or learning.

Formative assessment vs. summative assessment

Formative assessment improves learning in progress. Summative assessment commonly evaluates learning at the end of a unit, course, or other instructional period. The distinction is mainly about purpose, timing, and use rather than a fixed set of task types.

Carnegie Mellon University describes the formative goal as monitoring learning to provide ongoing feedback. It describes the summative goal as evaluating learning against a standard or benchmark at the end of instruction. Yale’s Poorvu Center makes a similar distinction between monitoring progress during learning and evaluating proficiency at its conclusion.

The same task can serve different purposes. A quiz used only to produce a final score functions summatively. A similar quiz can function formatively if you examine the responses, identify patterns, provide feedback, and give students an opportunity to apply that feedback.

The categories can also overlap across time. Carnegie Mellon notes that information from a summative assessment can be used formatively when it guides work in a later course or learning period. A final unit assessment remains summative for that completed unit, but its item patterns might shape review, curriculum planning, or support during the next one.

A graded activity can therefore contribute to a formative process, but high consequences can work against the purpose of practice. NWEA’s guidance says evidence gathered during practice should inform instructional decisions and advises against using formative-practice scores for grading or placement. A practical approach is to keep checks low stakes, use points sparingly if required, and follow your school’s grading policy without turning an unfinished learning attempt into a high-consequence judgment.

The formative assessment cycle

A classroom check becomes formative when it connects a learning goal to a response. That response then changes what the teacher or student does. NWEA describes a process of clarifying goals, eliciting and analyzing student thinking, and adjusting learning strategies or instructional steps.

The cycle has five connected parts:

1. Clarify the learning goal and what successful work would show.

2. Elicit evidence that is relevant to that goal.

3. Interpret what student responses reveal.

4. Give feedback that identifies a workable next step.

5. Adjust teaching or give students time to respond.

Collecting answers completes only the evidence-gathering part. If the answers are recorded but do not influence feedback, student work, or instruction, the formative cycle remains unfinished.

Start with a clear learning goal and useful evidence

Begin by identifying what students should understand or be able to do. Then state the success criteria, the observable qualities that would indicate progress toward that goal. Both the Iowa Department of Education and NWEA place clear learning goals near the beginning of effective formative practice.

The goal determines what kind of response you need. If students are learning factual recall, a brief written response may be enough. If the target is mathematical reasoning, a final answer alone is too shallow. You may need a documented solution, explanation, diagram, or think-aloud that exposes how the student reached the result.

Success criteria also give you something specific to inspect. “Understand fractions” is too broad to guide a useful check. A narrower goal might ask students to justify why two fractions are equivalent using a representation. The response should then include the representation and justification, not only a selected answer.

Participation is not the same as evidence of learning. A lively poll can tell you that students responded, but its usefulness depends on the question and available answer choices. Before choosing a technique, finish this sentence: “This response will help me see whether students can…” If the sentence remains vague, clarify the learning goal or change the task.

The evidence does not need to be extensive. It needs to be relevant enough to distinguish among plausible next moves. A short explanation that reveals a misconception can be more instructionally useful than a page of answers that shows only correct or incorrect results.

Interpret response patterns and choose the next move

Interpretation begins by grouping responses according to what they reveal, not simply sorting students into correct and incorrect groups. NWEA emphasizes eliciting and analyzing evidence of student thinking before choosing the next instructional step.

Look for the breadth and type of the pattern:

• A widespread shared misconception may call for whole-class clarification or a different representation.
• A misconception concentrated among some students may call for targeted practice in a small group.
• An isolated error may be better handled through individual feedback.
• Secure understanding may support enrichment, a transfer task, or moving forward.
• Mixed or ambiguous evidence may call for another focused question before making a larger change.

There is no universal percentage that automatically means “reteach” or “move on.” The decision changes with the learning goal, the type of error, the consequences of moving forward, and any other evidence you have.

A foundational misunderstanding deserves attention because later work may rely on it. A minor notation mistake may not justify stopping the whole class if the student’s reasoning is otherwise sound. Likewise, one multiple-choice response may be too ambiguous to distinguish a misconception from a misread question or accidental selection.

Plan possible responses before collecting the evidence. You do not need to predict every student answer, but you can identify the most important branches. For example: if students confuse two concepts, compare them directly; if they understand the concept but make procedural errors, provide focused practice; if they can explain and apply the idea, move to a less familiar context.

This advance planning keeps formative assessment manageable. You are collecting information because it can change a decision, not because every available data point must be recorded.

Turn feedback into student action

Useful formative feedback tells students where their work currently stands, what needs attention, and what they can do next. It is specific, connected to the success criteria, and forward-looking.

Yale’s Poorvu Center recommends feedback tied to predefined criteria, followed by an opportunity to revise or apply it before final submission. NWEA similarly emphasizes giving students both the time and the process needed to use feedback.

“Check your work” gives a student little direction. “Your representation shows equal parts, but your explanation does not connect those parts to the numerator” points to a specific gap. The next task might ask the student to add that connection, compare the work with an example, or apply the same reasoning to a new fraction.

Timing matters because feedback must arrive while students can still act on it. A detailed comment received after the class has moved permanently to another topic may explain a past result, but it has less opportunity to shape the learning that produced it.

Student self-assessment and peer feedback can also generate formative evidence. Yale recommends asking students to use course criteria to evaluate their own or a peer’s work. When students identify which criterion has been met, cite evidence from the work, and name a next step, they are doing more than participating. They are making their understanding of quality visible.

Peer feedback needs a clear focus. Asking students to “give feedback” can produce praise, editing, or personal preference. Asking them to check whether a solution explains each step, then identify one unsupported transition, directs attention to the learning target. The teacher can inspect those comments as evidence of both the original work and the reviewers’ understanding.

Choose a formative assessment technique by timing and purpose

Choose a technique according to when you need information, what students must demonstrate, and what decision the response could change. A long menu of formative assessment examples is less useful than a clear match between the technique and its intended evidence.

The Iowa Department of Education lists approaches including pretests, focused questions, concept maps, written assignments, surveys, interviews, and exit tickets. Think-alouds expose a student’s verbal reasoning, while documented solutions record the steps used to solve a problem. Exit tickets provide a brief end-of-lesson check.

These techniques can be technology-free or digital. A poll might use a show of hands, response cards, mini whiteboards, or a digital response system. A documented solution might be written on paper or submitted electronically. The delivery method matters less than whether the response captures the intended learning and arrives in time to guide action.

Fast techniques and deep techniques serve different decisions. A visual signal can tell you that students feel ready, confused, or undecided. It usually cannot show the reasoning behind that judgment. A think-aloud or documented solution takes more time, but it can reveal the step where understanding breaks down.

You can combine techniques when the first response is too shallow. Start with a class poll, then ask selected students to explain their choice. Use an exit ticket to identify a pattern, then open the next lesson with a focused question that tests whether the clarification worked.

No technique fits every learner, subject, or goal. The strongest choice is the smallest task that produces enough relevant evidence to support your next decision.

Check that the response reveals the intended learning

A suitable response format makes the target visible without adding unrelated demands that obscure it. Before using a technique, identify exactly what the response will demonstrate.

A confidence rating answers “How sure is the student?” It does not directly answer “Can the student explain, perform, or apply this skill?” Confidence still has a use. A student who is correct but uncertain may need a different next step from a student who is correct and can justify the reasoning. Pairing confidence with a brief demonstration keeps those two kinds of information separate.

Check the fit between the task and the target:

• Does the response directly demonstrate the knowledge, reasoning, or performance in the learning goal?
• Could reading, language, handwriting, motor, sensory, or technology demands interfere with that demonstration?
• Can students respond in another format without changing what is being assessed?
• Will the resulting work help you distinguish among realistic next actions?

For example, if the goal is to explain mathematical reasoning, an oral explanation, diagram, typed response, or handwritten solution might all provide relevant evidence. They are not automatically interchangeable. The appropriate option depends on which features of the response belong to the learning target.

If handwriting quality is not part of the goal, difficulty producing neat written work should not become the deciding evidence. If precise mathematical notation is part of the goal, changing the response to an entirely verbal format may remove something you need to assess. Accessibility is not simply making the task easier. It is reducing unrelated barriers while preserving the intended learning target.

Digital response tools can improve speed or make some response formats easier to collect. They can also introduce device access, navigation, or input demands. A paper response can remove those demands, but it may introduce different barriers. Choose the format by comparing its demands with the construct you want to see, meaning the specific knowledge or skill being assessed.

From student work to the next instructional move

Consider an illustrative math lesson with this learning goal: students will solve a linear equation and justify each transformation by referring to the equality of both sides. The teacher needs to inspect reasoning, so students submit documented solutions rather than final answers alone.

Assume the teacher has already shared two success criteria. Each transformation must preserve equality, and the written work must make the operation on both sides visible. The purpose of the task is not to calculate a final grade. It is to decide what students need before moving to a more complex equation.

Documented solutions are useful here because students record each step they took. A correct final answer can then be checked against the reasoning that produced it. An incorrect answer can be examined for the exact point where the approach changed.

Suppose one response pattern shows students applying an operation to only one side of the equation. If that pattern appears across much of the class, a whole-class clarification may be appropriate. The teacher might return to a balance representation, contrast valid and invalid transformations, and ask students to correct one example before continuing.

A second pattern may show that some students preserve equality correctly but make errors when combining signed numbers. Those students do not necessarily need the full concept retaught. A focused group task on signed-number operations may address the narrower obstacle while the rest of the class continues with equation reasoning.

An individual student might use a valid alternative sequence of transformations but omit the explanation for one step. Individual feedback can point to the missing justification without treating the entire solution as incorrect. The student can revise that step and demonstrate that the success criteria are now met.

Students whose solutions are accurate, justified, and consistent can move to enrichment. They might compare two valid solution paths, decide which is more efficient for a particular equation, or analyze a deliberately flawed solution. The extension should deepen the same learning goal rather than supply extra work for finishing early.

The class may also produce responses that are too brief to interpret. If several papers contain only final answers, the teacher cannot tell whether students understand equality, memorized a procedure, or guessed. The next move is not automatically reteaching. It may be another prompt that requires one transformation and a written justification.

This example shows why counting incorrect answers is not enough. Two sets of student work can contain the same number of incorrect final answers yet require different responses. One set may reveal a shared conceptual misconception suitable for whole-class attention. Another may contain unrelated slips, incomplete explanations, and one prerequisite gap that call for targeted responses.

The decision-changing evidence is the reasoning pattern. That pattern helps determine whether to clarify for everyone, support a group, respond to an individual, offer enrichment, collect better evidence, or continue.

Common formative assessment failures and how to prevent them

Formative assessment breaks down when a check collects activity without producing useful evidence or action. The practical correction is to tighten the connection among the learning goal, student response, interpretation, and next step.

• The learning goal is unclear. Define what students should know or do, then identify the qualities you expect to see in their work.
• The evidence is too shallow. If a final answer cannot reveal the reasoning you need, ask for a step, explanation, representation, comparison, or performance.
• Confidence is mistaken for understanding. Use confidence signals to identify students’ perceptions, then pair them with a task that directly demonstrates the target.
• Too much evidence is collected. Gather only what can inform a realistic decision. A focused sample may be more useful than a large set of responses you cannot inspect in time.
• Feedback arrives too late. Build the check early enough for feedback to shape revision, practice, or the next lesson.
• Students cannot apply the feedback. Include a correction, revision, retry, explanation, or new application. Feedback without an opportunity to respond leaves the learning step incomplete.
• Instruction continues unchanged. Decide in advance which patterns would lead to whole-class clarification, targeted support, individual feedback, enrichment, or continuation.

These failures often occur even when the classroom activity looks appropriate. An exit ticket, for example, is not automatically formative. It becomes formative when the teacher examines the responses and uses them to shape what happens next.

Workload is part of the design decision. A technique that produces more information than you can interpret promptly may delay feedback until it loses instructional value. Shorter responses, sampling, carefully chosen questions, or grouping by response pattern can keep attention on consequential evidence.

The correction is not to collect data continuously. It is to choose moments when information can change a decision. A focused question before independent practice may prevent a misconception from spreading. An exit ticket may help plan the next lesson. A documented solution may be worthwhile when you need to see reasoning that a quick poll cannot reveal.

What research says about benefits and limitations

Research supports formative assessment as a promising K–12 instructional practice, but it does not identify one optimal strategy for every classroom. Reported effects vary by the type of formative assessment, and the certainty of the available research is mostly low or very low.

A 2024 umbrella review published in Sustainability examined 13 meta-analyses of formative-assessment interventions in primary and secondary education. An umbrella review synthesizes findings from multiple meta-analyses rather than combining individual classroom studies directly.

The review reported effects ranging from trivial to large and found that effect magnitude varied by formative-assessment type. The included work concentrated mainly on cognitive learning outcomes and, to a lesser extent, affective outcomes.

Methodological quality and certainty are separate issues. Of the 13 included meta-analyses, the review rated ten as moderate quality, one as high quality, and two as low quality using AMSTAR-2. Its GRADE assessment rated the robustness of the evidence as very low for nine, low for three, and moderate for one. On that basis, the authors concluded that the low-to-very-low certainty prevented robust recommendations about the optimal formative-assessment strategy for K–12 learning.

The review’s broad summary described positive effects and said no negative effects were identified. Its fuller results also reported a trivial negative effect in one subgroup involving curriculum-embedded, open-ended formative assessment schemes. These statements can coexist because an overall summary can be positive while a narrower subgroup estimate differs. The practical conclusion is not that every implementation helps every learner. Results vary with the assessment type and how the process is carried out.

The process matters. Clear goals help teachers choose relevant evidence. Specific, criteria-linked feedback gives students a path forward. Time to revise or apply feedback turns comments into learning activity. Analysis of response patterns helps teachers adjust instruction rather than merely record performance. These components are central to the guidance from Yale, NWEA, and the Iowa Department of Education.

“Feedback” alone does not describe its quality or usefulness. A score with no improvement path, a comment that arrives after the opportunity to revise, and specific guidance followed by another attempt are all different classroom experiences. Treating them as equivalent hides the mechanism that formative assessment depends on.

The most defensible classroom approach is therefore not to search for a universally superior technique. Start with a clear learning goal. Choose a response that reveals the intended knowledge or reasoning. Decide how different patterns could change instruction. Then make sure students have a meaningful opportunity to use the resulting feedback.