Formative assessment: what it is, how to do it, and what the research says

What it is

Formative assessment is any check on learning whose main job is to change what happens next. A teacher asks a question, every student answers on a mini whiteboard, and the teacher sees that half the room has the wrong idea: that's formative assessment, as long as the next ten minutes change because of it. The same quiz, marked and filed, would be summative: a measure of what was learned, used to report rather than to steer.

The name describes a use, not a kind of task. Exit tickets, hinge questions, mini whiteboards, thumbs up, quick writes, conferring with a student, a quiz, a draft with comments: any of these can be formative, and none of them is formative if nothing changes afterward.

One practical way to organize it comes from Dylan Wiliam and Siobhán Leahy's guide for teachers, which sets out five strategies:

1. Make the goal clear. Share what students are learning and what success looks like (success criteria), often with examples of good and weaker work.
2. Bring out evidence of learning. Plan questions, discussions and tasks that show you what students actually think, not just whether the fast hands know.
3. Give feedback that moves learning forward. Comments that tell the student what to do next, and time to do it.
4. Make students resources for each other. Peer feedback and talk, structured so it's useful.
5. Make students owners of their own learning. Self-assessment against the success criteria, and students choosing what to work on next.

Formative assessment overlaps with retrieval practice: a warm-up quiz both strengthens memory and shows you what to reteach. The difference is the purpose. Retrieval practice is for the student's memory; formative assessment is for your next decision.

Why it works

Teaching without checking is teaching on assumptions. Students can sit quietly through an explanation and leave with a misconception you'd fix in two minutes if you knew about it. Formative assessment closes that loop while there's still time to act: in the same lesson, the next day, or before the unit test rather than after.

Paul Black and Dylan Wiliam's 1998 article "Inside the Black Box" argued, from their review of the research, that formative assessment is an essential part of classroom work and that improving it can raise achievement.

Later research has confirmed the direction and shrunk the size. Kingston and Nash screened over 300 K–12 studies, found only 13 that gave usable effect sizes, and estimated a modest average effect, well below the figures often quoted. Lee and colleagues reviewed 33 US K–12 studies with control groups and found a small positive effect, with similar benefits in math, literacy and the arts. A What Works Clearinghouse-style review for elementary schools by Klute and colleagues found a positive effect on achievement overall, larger in math than in reading and writing.

The strongest single test is a randomized trial in 140 secondary schools in England, reported by Anders and colleagues. Teachers met in school-based learning communities to build formative assessment into ordinary lessons, and pupils' national exam results rose by a small amount. That's a modest effect, but it was measured on national exams marked outside the schools, across 140 schools, after light-touch training.

How to start, by grade

Examples

• The hinge questionGrades 3–12

  Halfway through a fractions lesson: "Which is bigger, 3/8 or 3/5? A: 3/8, because 8 is bigger. B: 3/5, because fifths are bigger pieces. C: They're the same, because the tops match." Everyone holds up a letter. If most choose B, move on. If many choose A, you know exactly which idea to fix, and you fix it now.

• Three-pile exit ticketsGrades K–8

  Each student answers one question on a slip as they leave. After school you sort the slips into three piles: got it, nearly, not yet. Tomorrow the not-yet group works with you for ten minutes while the others start practice. It takes five minutes to sort a class set, and it changes the next lesson.

• Mini whiteboards, all at onceGrades 1–12

  Ask a question, give thinking time, then say "Show me." Every student holds up an answer at the same moment, so nobody copies and you see the whole room in two seconds. Scan for patterns, not for the right answer: three students with the same wrong answer usually share the same wrong idea.

• Success criteria with work samplesGrades 3–12

  Before students write, show two anonymous paragraphs from last year (or ones you wrote). Ask which is stronger and why. Turn their reasons into a short checklist of success criteria, then use the same checklist for self-assessment, peer feedback and your comments.

• Comment first, grade laterGrades 4–12

  Return drafts with one strength and one next step tied to the success criteria, and no grade. Give fifteen minutes in class to act on the comment. Collect the improved version for the grade. Students read the comment because there's something to do with it.

Common mistakes

Collecting evidence and doing nothing with it.

An exit ticket that nobody reads, or a whiteboard check followed by the lesson you'd planned anyway, isn't formative. Decide in advance what you'll do with each likely result, even if it's only "reteach this example tomorrow".

Only checking the volunteers.

Hands-up questions show you what the confident students think. Use all-student responses (whiteboards, fingers, slips) and random calling so the evidence covers the whole class.

Grading everything formative.

When every check counts toward a grade, students hide what they don't know. Keep formative checks low-stakes and separate from the gradebook.

Feedback without time to use it.

Comments on finished work that's already graded are mostly ignored. Give feedback while the work can still change, and build class time to act on it.

Treating it as a product to buy.

Software can help, and Kingston and Nash found larger effects for computer-based systems and for professional development, but the core is a teacher's habit of asking, looking and adjusting. Start with paper and a plan.

Trying all five strategies at once.

Pick one technique, such as hinge questions or exit tickets sorted into piles, and use it daily for a month before adding another. The school trial in England worked through teacher-led learning communities inside schools, not a single training day.

Time and cost

Time in class: two to ten minutes a lesson for checks; more on days you regroup students.

Time outside class: a few minutes to read and sort exit tickets, and planning time to write good hinge questions. The planning gets faster as you build a bank of questions for each unit.

Money: close to nothing. Mini whiteboards and markers, paper slips, or the free tools below.

Training: this is where the real cost is. Bennett's critical review points out that doing it well takes teachers substantial time and support. The trial in England worked through teacher-led learning communities inside each school, not a one-off course.

Debates and open questions

How big is the effect, really? You'll often see effect sizes of 0.4 to 0.7 quoted for formative assessment. Kingston and Nash found them unsupported by the studies they could use, and Bennett called the commonly quoted numbers suspect because they come from sources that are flawed, dated, unpublished or untraceable. Later reviews (Lee and colleagues; Klute and colleagues) and the English trial all find positive effects, but smaller ones. The practice is worth doing; the famous number isn't worth quoting.

What counts as formative assessment? Bennett argues the term covers so many different practices that results should be expected to vary widely. A study of computer-adaptive quizzes and a study of teacher questioning may both be filed under formative assessment. That's one reason effect sizes vary so much between studies.

Should students assess themselves? Here two reviews point different ways. Lee and colleagues found larger effects when students were supported to assess their own work. Klute and colleagues, looking only at elementary studies, found larger effects when a teacher or a computer program directed the assessment, and in reading found teacher-directed forms worked better than student-directed ones. A reasonable reading: self-assessment helps when students are taught how to do it against clear criteria, and younger students need more of the teacher's direction.

Which subjects? Kingston and Nash found larger effects in English language arts than in math or science; Klute and colleagues found larger effects in math than in reading and writing. With so few rigorous studies in each subject, we wouldn't choose a subject to start in on the basis of either review.

Tools that do it

Free, on this site, nothing to sign up for. Class lists stay in your browser.

• Exit tickets: printable slips for a one-question check at the end of a lesson; sort them into three piles after school.
• 3-2-1 exit slip: three things learned, two questions, one thing to remember: shows you what stuck and what didn't.
• Exit ticket wheel: spin for an end-of-lesson prompt when you haven't planned one.
• Random name picker: no-hands questioning: everyone thinks, then you pick names at random so the evidence isn't only from volunteers.
• Classroom quiz show maker: multiple-choice questions on the board, good for hinge questions with every student answering.
• Rubric maker: turn your success criteria into a rubric students can use to check their own work.
• I can goal sheet: students write the learning goal in their own words and check it off when they can do it.
• 25 formative assessment examples: quick checks you can use tomorrow, by grade.

Questions teachers ask

What's the difference between formative and summative assessment?

The use, not the task. Formative assessment happens during learning and changes what you teach next. Summative assessment comes at the end and reports what was learned. The same quiz can be either, depending on what you do with the results.

Do exit tickets count as formative assessment?

Only if you use them. An exit ticket you read, sort and act on the next day is formative assessment. A stack of exit tickets in a drawer is paperwork. Keep them to one or two questions so you can read a class set in five minutes.

Is the 0.4 to 0.7 effect size real?

Not as far as the research we could check shows. Kingston and Nash found those figures unsupported and estimated a much smaller average, and Bennett traced the big numbers to flawed or untraceable sources. Later reviews and a large trial still find positive effects, just smaller ones.

Should formative assessment be graded?

Generally no. Its job is to show what students don't know yet, and grades give students a reason to hide that. Record what you learn from it in your planning, not in the gradebook.

How do I find time for this?

Replace rather than add. Swap a hands-up question for a whiteboard check, swap marking finished work for commenting on drafts, and swap the end of a lesson for an exit ticket. Most techniques take the same minutes you already spend, used differently.

Does technology make it work better?

Kingston and Nash found larger average effects for computer-based formative systems and for programs built around professional development, but from a small number of studies. Technology helps you collect answers quickly; the teacher still has to decide what to do with them.