Training · Cross-cuttingMoveprovenverified

Formative Assessment Checkpoints

also known as formative assessment, assessment for learning, check for understanding, progress checkpoints

Tags: formative-assessmentBlack-Wiliamfeedback-loopself-assessmentcheck-for-understanding

Formative assessment is a continuous, embedded practice of gathering evidence about learner understanding and adjusting instruction — or the learner's own effort — in response to that evidence. Unlike summative assessment, which judges what was learned after the fact, formative assessment intervenes while learning is still in progress. Black and Wiliam's landmark 1998 review found strong evidence that improving formative assessment raises achievement standards, and that the feedback loop must close in both directions: the teacher adjusting instruction and the learner adjusting their own approach through self-assessment. Checkpoints make the practice visible and regular rather than implicit and occasional.

How the learner advances

Intent. Keep learning on track by regularly surfacing the gap between current understanding and the target, then using that gap information to adjust instruction or learner effort before it becomes a terminal failure.

When to apply. Apply throughout any learning sequence — not just at the end. Insert checkpoints wherever the learning sequence branches or depends on a prior concept being solid: if the prior concept is shaky, subsequent instruction builds on sand. Use brief, low-stakes checks — exit tickets, targeted questions, short tasks — rather than full tests, so the checkpoint itself does not consume disproportionate time. Apply especially when the instructor suspects misconceptions have formed, when learner engagement has dropped, or when the pace of instruction has been faster than normal. Train learners in self-assessment as early as possible so they can initiate their own checkpoints without waiting for external prompting.

Threshold — earns the next step. The learner can accurately self-assess their own understanding of the current topic — identifying what they know confidently, what they know partially, and what they do not know — and can initiate a request for targeted feedback on the specific gap.

Masterpiece — the artifact that proves it. A learning sequence in which every checkpoint generated a concrete instructional adjustment, the learner's self-assessments were consistently accurate against external evidence, and the final summative assessment showed substantially higher performance than a comparable sequence without checkpoints.

Facets

  • Container — async
  • Mode — adaptivefeedback-loopcoached
  • Reach — individual
  • Persona — instructorlearner
  • Craft (AI Fluency) — discernmentdiligence
  • Learner — humanautonomous-agent
  • Trainer — humanautonomous-agent

Inputs

  • Learning objective in checkable form — A specific, observable outcome that can be assessed in a short task or question — not 'understands recursion' but 'can write a recursive function that correctly handles the base case.'
  • Low-stakes checkpoint mechanism — A quick probe of understanding — question, brief task, exit ticket, peer explanation, or diagnostic quiz — that generates evidence without the stress and time cost of formal evaluation.
  • Commitment to act on the evidence — A willingness from the instructor or learner to actually change what happens next based on checkpoint results, rather than treating the checkpoint as a box-ticking exercise.

Outputs

  • More capable learner — A learner whose misconceptions were caught and corrected while the material was still current, rather than discovered retrospectively in a final assessment when remediation is harder and more costly.
  • Adjusted instructional path — The masterpiece of a checkpoint cycle: a concrete change to what happens next — reteaching a concept, slowing pace, giving targeted feedback, redirecting a learner's self-study — that demonstrably closes the gap identified.

Steps (5)

  1. Specify the checkpoint objective

    Before the checkpoint, articulate exactly what evidence of understanding you need: which concept, which operation, which misconception you are probing. Vague objectives produce checkpoints that generate noise rather than signal.

  2. Design a minimal, targeted probe

    Create the smallest checkpoint that generates reliable evidence of the specific objective. A single well-chosen question often outperforms a ten-item quiz in revealing whether a concept is understood or misunderstood.

  3. Administer and collect evidence

    Run the checkpoint in a low-stakes mode — no grades, no judgment — so learners respond honestly. Anonymous or group responses (show-of-hands, shared whiteboards, polling tools) reduce social pressure that masks real understanding.

  4. Interpret the evidence and decide

    Categorise responses: who understands, who has a specific misconception, who is guessing. Make a concrete decision about what changes: reteach, pair stronger with weaker, send one group ahead, slow down. The checkpoint is useless without this decision.

  5. Act and close the loop with the learner

    Implement the change. Crucially, give learners feedback on their checkpoint response — not a grade but a specific note on what was correct, what was wrong, and what to do next. Black and Wiliam are explicit: feedback must tell the learner how to improve, not just whether they succeeded.

Principles

  • Formative assessment works only when the loop closes: evidence gathered but not acted upon is surveillance, not assessment. The loop must close for both instructor (adjusting teaching) and learner (adjusting effort and strategy).
  • Feedback should be task-referenced, not person-referenced: telling a learner their answer was wrong is far less useful than telling them exactly which step went wrong and what to do instead. Grade-based or comparative feedback provokes ego responses that block learning.
  • Learner self-assessment is not a nice-to-have: training learners to assess their own understanding is what makes formative assessment scalable and persistent. A learner who can identify their own gap does not need an instructor at every checkpoint.

Known uses (2)

Known failure modes (3)

  • [checkpoint-without-adjustment]

    Anti-pattern: checkpoints are run on schedule but instruction does not change based on results. The checkpoint becomes a ritual with no feedback loop — learners sense the disconnect and stop engaging honestly with the probe.

  • [grade-as-feedback]

    Anti-pattern: the only feedback returned to the learner is a score or grade. Black and Wiliam's research is explicit that grades without task-specific guidance on improvement do not improve learning and can actively harm motivation.

  • [checkpoint-overload]

    Anti-pattern: so many checkpoints are inserted that assessment time crowds out learning time, and learners experience checkpoint fatigue. Formative checkpoints must be minimal — their purpose is to generate a decision signal, not to be a comprehensive measure.

Related trainings (5)

Sources (3)

Provenance

  • Ecosystem: education, professional development, e-learning, autonomous-agent training
  • Added to catalog:
  • Last updated:
  • Verification status: verified