A teacher asks a class to explain why a plant grew differently in two pots. Several pupils name sunlight; others mention water; a few repeat a phrase from the worksheet without connecting it to either pot. The teacher can move on, record a score, or use those answers to change the next ten minutes of teaching. Formative assessment is the third possibility: evidence of learning is elicited, interpreted, and used while there is still time to improve the learning. Its value lies in the decision made from the evidence, not in the mere presence of a quiz.
Assessment becomes formative when it changes the next step
A question, discussion, draft, demonstration, exit ticket, or short test can all serve a formative purpose. None is formative by format alone. If the teacher collects answers and files them without changing instruction, the activity may be measurement but it has not completed a formative cycle. If a learner uses the result to revise a strategy, or a teacher uses it to reteach a misconception, the same activity can support learning. This is a functional definition, not a claim that one tool always works.
Black and Wiliam’s influential review drew attention to classroom assessment that informs teaching and learning. Their later theoretical account put the emphasis on evidence about where learners are, where they are going, and how to move forward. The framework includes teachers, peers, and learners as agents. It is more demanding than adding frequent tests: someone must interpret an answer against a learning goal and decide what to do differently.
Summative and formative purposes need not be enemies. A final examination primarily certifies what was learned at a point in time. A practice question before that point may help locate a misunderstanding. The same task can yield both kinds of information, but the stakes and timing affect how honestly learners reveal uncertainty. When every tentative answer feels like a permanent grade, students may hide what they do not know. A formative environment needs room for errors to become usable evidence.
The cycle is easy to describe but difficult to execute well. A teacher must choose a question that reveals the relevant idea, hear or inspect enough responses, interpret their meaning, select a response, and check whether it helped. A correct answer can conceal a memorized procedure; a wrong answer can reveal a productive partial understanding. Good assessment therefore depends on knowledge of the subject and likely misconceptions, not only on efficient data collection.
What counts as useful evidence
Evidence is useful when it speaks to a decision. If the goal is to compare causal explanations, a multiple-choice recall item may say little about how students reason. If the goal is accurate retrieval of essential vocabulary, a short recall prompt may be exactly right. The assessment should sample the kind of performance the lesson is trying to improve. Otherwise, a neat dashboard can invite precise action on the wrong target.
A single response is often ambiguous. A pupil who hesitates may lack knowledge, misunderstand wording, need more time, or be reluctant to speak publicly. A teacher can ask for an explanation, inspect a written step, or compare responses across tasks. More data are not automatically better; the aim is enough evidence to distinguish plausible interpretations. Formative judgment is a disciplined inference, not a direct window into a learner’s mind.
Questions that invite all learners to respond can reduce the bias of hearing only confident volunteers. A brief written response or paired explanation can reveal a wider spread of ideas. Yet these methods have costs. Collecting too many answers can leave no time to act on them, and public display of responses can embarrass learners. The design should match class size, time, privacy, and the decision the teacher can realistically make next.
Learning goals and criteria make evidence interpretable. “Understand photosynthesis” is too broad for a ten-minute decision. “Explain why a plant with limited light grows differently when water is held constant” gives teacher and learner a more concrete target. Criteria should describe qualities of an explanation, not merely a checklist of words. A learner needs to see what would count as a stronger answer and why.
Evidence is only the start of a response
Suppose half the class confuses correlation with causation. The teacher might model a contrasting example, ask pairs to critique two explanations, or revisit the data before continuing. The choice should reflect the misconception and the remaining time. Repeating the same explanation louder may not help if the original representation was the problem. The formative move is an adjustment guided by the evidence, followed by a fresh check.
Black and Wiliam’s 2009 account frames formative assessment as a process in which information about learning is used to adapt teaching and learning activities. This avoids treating feedback as a one-way message from teacher to student. A learner may notice that a strategy failed and choose another; peers may compare reasoning and reveal a gap. The teacher still carries responsibility for task design, interpretation, and equitable participation.
Feedback should tell a learner something they can use. A score may signal distance from a goal but rarely explains how to close it. “Your answer names light and water, but does not say which changed between the pots” gives a next step. The learner also needs an opportunity to revise; otherwise the comment arrives as a postscript to work already abandoned. The connection between assessment and action is what makes the information formative.
A useful response may sometimes be to change the task rather than tell the learner the answer. If pupils can identify variables but cannot connect them to a comparison, the teacher might provide a partially completed table or ask them to predict an outcome before showing data. These are instructional choices, not universal recipes. The measure of success is whether the next attempt shows a more accurate or flexible understanding.
What reviews support and what they complicate
Black and Wiliam’s 1998 review helped establish formative assessment as a serious educational question, but it synthesized diverse studies and practices. It should not be read as a single randomized trial of one uniform method. Kingston and Nash later conducted a meta-analysis and called for more rigorous research. Their work illustrates why estimates vary: interventions differ in duration, subject, grade, comparison condition, and what researchers mean by formative assessment.
A 2022 meta-analysis focused specifically on K–12 reading achievement. It found variation across studies and reported that smaller studies tended to show larger effects. It also examined whether assessment was integrated with differentiated instruction and whether teachers and students both participated. These findings are relevant to reading interventions under the review’s inclusion criteria. They do not guarantee that the same features will work identically in mathematics, adult education, or another culture.
The reading review’s methodological cautions matter as much as its positive average. A small, intensively supported study may perform differently from a district-wide rollout. Some interventions bundle assessment with new materials, training, or technology, making it hard to isolate one ingredient. Outcomes may be measured by tests closely aligned to the intervention or by broader assessments. A claim that “formative assessment works” should identify what was done, for whom, and against what alternative.
Hattie and Timperley’s synthesis of feedback research helps explain why information can be powerful yet variable. Feedback can address the task, the process used to complete it, or self-regulation. Vague praise or criticism of the person may draw attention away from the work. This framework informs formative assessment, but feedback is not the whole cycle. A comment can be excellent in wording and still fail if the learner cannot act on it or the teacher does not see the result.
Making learners participants
A learner can use a worked example, a rubric, or a peer explanation to compare current work with a goal. Self-assessment is strongest when it names evidence: which part of an argument is supported, which calculation step is uncertain, what strategy was tried. Asking “How do you feel about your work?” can open a conversation, but it does not by itself reveal whether the concept is understood. Calibration needs examples and opportunities to revise.
Peer assessment can widen the sources of feedback. Two students may recognize different strengths in a draft, and explaining a criterion to a peer can sharpen one’s own understanding. Yet peers can also repeat misconceptions or turn comments into social judgment. The teacher needs to model what a specific, respectful, task-focused comment looks like and inspect whether the exchange helps both parties. Peer work is not a way to outsource professional judgment.
Students should know that a formative question is an invitation to expose thinking, not a trap. If errors are punished immediately, the data become less informative. This does not require pretending all answers are equally good. It requires treating an error as a clue about the next instructional move while preserving clear standards. A learner can be held to a demanding goal and still be given safe opportunities to show the gap.
Accessibility matters. A timed written response can underrepresent a learner who understands the concept but needs more processing or a different mode of expression. An oral explanation may disadvantage someone who needs time to formulate language. Multiple routes to show thinking can improve interpretation, provided they still test the intended skill. The question is whether the assessment is revealing the target learning or a barrier in the format.
Digital tools and the temptation of easy data
Polling software can quickly show a distribution of answers; adaptive practice can flag repeated errors. These tools may help a teacher notice patterns in a large group. They do not interpret every misconception or decide the best next lesson. A dashboard that classifies a student as “red” can be useful for triage, but it can also obscure why the answer was wrong. The teacher must be able to inspect the underlying task and response.
Automated feedback can be timely. It is particularly suited to some constrained questions with clear criteria, such as a calculation or factual retrieval. Complex explanation, creativity, and argument require more careful interpretation. A fluent AI comment may sound specific while misreading a student’s intention or inventing a criterion. Any digital system should be tested against actual student work, with attention to errors across groups and opportunities for challenge.
Data volume can reverse the purpose of assessment. If teachers spend the lesson tagging responses but have no time to adapt instruction, collection has displaced action. A smaller sample of well-chosen responses may be more useful than a large set of poorly aligned metrics. The design question is not “How much can we measure?” but “Which observation can change the next decision?”
Privacy also shapes trust. Formative responses often reveal uncertainty and unfinished thinking. Learners should know who sees their work, how long it is kept, and whether it affects formal evaluation. A permanent record of every tentative answer can make exploration feel risky. The institution should collect only what serves a clear learning purpose and give teachers room to use informal evidence without turning each moment into surveillance.
Evaluating a formative routine
A teacher or team can evaluate a routine by tracing its full cycle. What question was asked? What range of responses appeared? What interpretation followed? What changed in teaching or practice? What happened on a later, independent task? This sequence is more informative than counting quizzes administered. It can reveal a weak question, a mistaken interpretation, an ineffective response, or an improvement that did not transfer.
Comparison matters. If learning improves over a term, maturation, extra practice, curriculum changes, or other support may contribute. A local improvement project can still be valuable without proving a causal effect, but claims should match the design. Larger evaluations can compare groups or time periods and examine whether implementation is consistent. Qualitative inspection of student work can explain why a numerical result changed or failed to change.
Equity should be part of evaluation from the start. Does every learner have a chance to show thinking? Are some groups more often interpreted as lacking ability when the task wording or classroom norms are the obstacle? Do follow-up opportunities reach those who need them? Aggregate gains can hide uneven benefit. Formative assessment promises responsiveness; it should be judged partly by whether the response is available fairly.
In adult learning, the same logic can apply without copying a school routine. A trainee can attempt a realistic task, receive evidence about a particular decision, try a revised approach, and check it again. The task and stakes differ from a classroom, so school effect sizes should not be imported directly. What transfers is the basic question: can evidence about current performance change the next opportunity to learn?
What the evidence does not show
The evidence does not show that frequent testing is automatically formative, that every feedback message improves learning, or that a single effect size applies across subjects, ages, and systems. Reviews combine heterogeneous practices; some evaluations are small or bundle assessment with other changes. A formative routine can fail if the question targets the wrong construct, the interpretation is mistaken, the response is unusable, or the learner has no chance to revise. Strong claims require a specified practice, population, comparison, and outcome.
The central standard remains operational. Elicit evidence of a meaningful learning goal, interpret it with humility, use it to choose a better next step, and check whether that step helped. This is more work than adding a score to a page, but it gives assessment a purpose while learning is still open to change.
Research references
- Black, P., & Wiliam, D. (1998). Assessment and classroom learning. Assessment in Education: Principles, Policy & Practice, 5(1), 7–74. View source
- Black, P., & Wiliam, D. (2009). Developing the theory of formative assessment. Educational Assessment, Evaluation and Accountability, 21, 5–31. View source
- Kingston, N., & Nash, B. (2011). Formative assessment: A meta-analysis and a call for research. Educational Measurement: Issues and Practice, 30, 28–37. View source
- Xuan, Q., Cheung, A., & Sun, D. (2022). The effectiveness of formative assessment for enhancing reading achievement in K–12 classrooms: A meta-analysis. Frontiers in Psychology, 13, 990196. View source
- Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–112. View source