Design Formative Assessment Strategy
Build a systematic formative assessment strategy that provides timely, actionable feedback to learners and instructors during learning, before high-stakes summative assessment.
Why This Is Best Practice
Adopted by: OECD educational reform recommendations, PISA framework, US Common Core assessment standards, all major national curriculum frameworks
Impact: Black & Wiliam (1998) meta-analysis of 250+ studies shows formative assessment produces effect sizes of 0.4–0.7 (equivalent to 2-grade improvement); Hattie (2009) ranks feedback as #1 influence on student achievement (ES = 0.73)
Why best: Formative assessment closes the gap between where the learner is and where they need to be — in real-time, during instruction, when it can still influence learning. Summative assessment after the learning period is too late to improve the outcome being measured.
Sources: Black & Wiliam "Inside the Black Box" Phi Delta Kappan (1998); Hattie "Visible Learning" (2009) Ch. 9; OECD "Formative Assessment" (2005)
Steps
- Define the learning intention — state clearly what learners are working toward in learner-facing language; formative assessment only works if learners understand the target they are being assessed against.
- Define success criteria — co-create or share explicit success criteria with learners: "You will know you have succeeded when you can X, Y, Z"; these become the reference point for all feedback.
- Select assessment techniques by purpose — checking understanding (exit tickets, mini-whiteboards, poll questions); diagnosing misconceptions (diagnostic questions, concept mapping); monitoring progress (performance tasks, portfolios); these serve different instructional decisions.
- Design diagnostic questions — create two-tier diagnostic questions: the answer + the reason; common wrong answers reveal specific misconceptions; this is more informative than right/wrong binary.
- Build in real-time checks — use: thumbs up/down/sideways, digital polls (Mentimeter, Poll Everywhere), 3-2-1 reflection (3 things learned, 2 questions, 1 application), cold calling with wait time, pair-share.
- Design exit tickets — 2–3 targeted questions at the end of each session that reveal whether the session's learning objective was achieved; review before next session to adjust instruction.
- Create feedback protocols — feedback must be: specific (what exactly is correct or needs revision), actionable (what should the learner do next), timely (within 24–48h of the task), and feed-forward (pointing toward the next step, not just evaluating the past).
- Teach learners to self-assess — share the rubric or success criteria before the task; require self-assessment before submission; learners who accurately self-assess outperform those who do not (Andrade 2010).
- Use peer assessment with structure — structured peer feedback (using a rubric or sentence stems) provides additional feedback without adding instructor workload; train learners in giving specific, constructive feedback.
- Act on assessment data — formative assessment that does not change instruction is administrative overhead; review exit ticket data before each session; modify pacing, re-teach, or extend based on evidence.
Rules
- Formative assessment data must be used to change instruction — if assessment data is collected but not acted upon, it is not formative, it is just frequent testing.
- Grades must not be assigned to formative assessments — grades turn formative tasks into summative ones and inhibit honest self-disclosure of confusion.
- Feedback quality beats feedback frequency — one piece of specific, actionable feedback outperforms five generic comments.
- Success criteria must be shared with learners before the task — criteria withheld until after the task cannot guide the learning process.
- Formative assessment requires psychological safety — learners must feel safe revealing misconceptions without penalty; grading formative work destroys this safety.
Common Mistakes
- Grading formative assessments — once graded, learners optimize for the grade rather than honest self-assessment; formative data becomes corrupted.
- Exit tickets without action — exit tickets stored but not reviewed before the next session produce no instructional benefit.
- Feedback that is purely evaluative — "This is good" or "Needs more work" is evaluation, not feedback; feedback must specify what and how.
- Using only teacher-assessment — peer and self-assessment are powerful formative tools; over-reliance on teacher assessment creates a single point of feedback bottleneck.
- Questions with only one correct pathway — diagnostic questions must have common incorrect answers; if everyone gets it right, you learn nothing about misconceptions.
When NOT to Use
- High-stakes summative assessment contexts (exams, certifications) where process does not affect scoring
- Very short learning interactions (<30 minutes) where overhead of structured assessment outweighs benefit
- Self-directed adult learning where the learner self-manages progress monitoring