Formative assessment
Assessment that exists to improve the work while it is still being made, as opposed to grading it once it is done. It counts as formative only when three things are true for the learner: the marks of good work are visible and understood, the learner can see how their current work measures up, and they can act to close the gap. The active part lives with the learner, not in the comment the teacher writes.
Assessment that exists to improve the work, in contrast with summative assessment that exists to judge the work after the fact. The technical content of the term, as Sadler gave it in 1989, is that an assessment is formative only if three conditions hold for the learner: the criteria for good work are visible and understood, the learner can see how the current work compares to those criteria, and the learner can act on the gap. A piece of feedback that satisfies only the first or second condition is not yet formative; what makes it formative is whether it closes a loop the learner can close.
Etymology§
The word formative in this sense was coined by Michael Scriven in The Methodology of Evaluation (1967) — a paper about programme evaluation that distinguished assessments designed to improve a programme as it was being built (formative) from assessments designed to judge it at completion (summative). Benjamin Bloom and colleagues carried the distinction into classroom assessment through the 1970s. The theoretical reformulation that made formative assessment a research field in its own right is Sadler's 1989 paper in Instructional Science, which specified the three-condition account and located the active component in what the learner could do with the feedback rather than in what the teacher gave. The construct entered mainstream UK education policy through Black and Wiliam's 1998 meta-review under the renamed banner assessment for learning.
The core distinction in the term is the one Scriven made in 1967: formative assessment exists to feed back into the work, summative assessment exists to grade it. The two are not different instruments; the same test can be either, depending on what follows. A weekly quiz used to identify what the class has not yet understood, and to reshape the next lesson, is formative. The same quiz, recorded as a contribution to the term grade, is summative.
Sadler's 1989 paper tightened what the term required. The earlier accounts described what the teacher was trying to do; Sadler described what would have to be true on the learner's end for the attempt to work. The three conditions — visible criteria, comparison to those criteria, action on the gap — locate the formative work inside the learner rather than inside the teacher. Feedback that the learner cannot act on is, by Sadler's definition, not formative; it may still be informative, accurate, or institutionally required, but it is not doing what the term names. After the shift, the research question stopped being was the feedback given and became was the gap closed.
The construct's policy reach came through Inside the Black Box in 1998. Paul Black and Dylan Wiliam reviewed over 250 studies of classroom assessment and reported large effect sizes (0.4–0.7 standard deviations) for formative practices implemented properly. The booklet, written for a non-specialist audience, was read by UK education policymakers and led to the Assessment Reform Group's Assessment for Learning programme through the early 2000s. The Assessment Reform Group used the phrase assessment for learning; the three-condition account is Sadler's.
The companion theoretical synthesis is Hattie and Timperley's 2007 review in Review of Educational Research, which classified feedback by what level it addressed — the task, the process of doing the task, the self-regulation behind it, or the person as a whole — and reported that the first three help and the last hurts. The empirical result that scored feedback (even when intended formatively) undermines the learning effect is Ruth Butler's 1988 study, which showed that students who received grades alongside comments paid attention to the grades and ignored the comments, whereas students who received only comments improved. The implication — that the assessment apparatus most schools run actively prevents the formative function from operating — is one of the seams of the topic and the point at which the assessment literature meets the ungrading argument.
The visible-criteria condition requires criteria that can be made visible — which is straightforward for a well-formed problem set and hard for an open-ended piece of writing. The gap-closing condition requires that the learner has time and permission to revise, which most institutional timetables do not grant. The construct's history through the 2000s and 2010s is substantially the history of attempts to fit it inside grading structures that were built for the summative function — with the ungrading and specifications-grading moves emerging when those attempts stalled.