Objective → module → practice → rubric row → fidelity metric. Every row must be fillable; an objective with no metric is decoration, and a metric with no objective is surveillance. Both move-name columns use the shared vocabulary from the Socratic AI Tutor’s pedagogy rules (v1.0, 2026-10-25).
Objective (observable) Module Practice Rubric row Fidelity metric Greet each student by name and reference their recorded work in the first 3 minutes M1 Session A round 3 (first-3-minutes ×3 reps)D1 levels 3–4 Rubric mean D1 ≥3.0; session-plan “name + true sentence” line Spread attention; every student produces at least once per session M1 Session A round 2 (fast-finisher card)D1 level 4 Observation tick “every student produces” Run the five-block 30-minute routine inside its time boxes M2 Session A rounds 1–2 (timed, full routine)D2 levels 3–4 Session logs; observed block timings Administer and collect the exit ticket every session — even if incomplete M2 Session A round 1D2 level 4 (ticket required) Sessions-with-ticket ≥90% (dashboard row)
Objective (observable) Module Practice Rubric row Fidelity metric Open every problem with a probing question before explaining (L0) M3 Session B round 1 · spot-the-move q1, q6D3 level 4 Rubric D3; observation tick “asked before explained” Escalate one ladder rung per turn (levels 0–5), hand on the rung M3 Session B rounds 1–3 (ladder cards)D3 level 4 Observation tick “one rung per turn” (rung sequence logged) Respond to an error with a probing question that exposes the misconception, never the label M3 Session B round 1 · spot-the-move q7, q8 · role-play card reveals D3 levels 3–4 Rubric D3; pre/post error-response task in the pilot (descriptive only) Hold 3–5 seconds of wait time after questions; never answer for the student M3 Session B round 2 (observer counts silence)D3 level 4 Observation tick “wait time 3–5s (counted)” After ~3 failures on a step: worked example of a similar problem (L4), then hand the original back M3 spot-the-move q3 · T2 annotations 8–10 D3 level 4 Observation tick “L4 worked example if 3 failures”
Objective (observable) Module Practice Rubric row Fidelity metric Give specific, process-focused feedback naming the work (no bare “good job”) M4 Session B round 3 · spot-the-move q5D4 levels 3–4 Rubric D4; observation tick “specific feedback naming the work” Confirm and ask why (once) after correct answers M4 Session B round 3 chain drill · spot-the-move q9 D4 level 4 Observation tick “confirm + one why + similar problem” Offer a similar problem after each solved problem M4 Session B round 3 chain drill · spot-the-move q10 D4 level 4 Observation tick (same chain)
Objective (observable) Module Practice Rubric row Fidelity metric Choose the exit ticket before the session; record results same day with tags M5 Filling the session plan template + tracker weekly D5 levels 3–4 Sessions-with-ticket ≥90%; tracker same-day check at observation Choose next session’s review and target from ticket results; write the link on the plan M5 Module 5 practice (four-ticket scenario) D5 level 4 Plan review at observation (“review topic named from ticket data”) Flag any student below mastery on the same skill tag two sessions running M5 Tracker flag exercise (L.R. case) D5 level 4 Flag rate and resolution time (dashboard) Keep records privacy-safe: initials only, no photos, org-managed storage M5 · safeguarding placeholder Tracker rules acknowledgment D5 (practice constraint) Data-privacy rules in the measurement plan
If a proposed training objective fills no metric row, it isn’t measured — cut it or measure it. If a fidelity metric fills no objective row, it isn’t trained — stop collecting it or start training it. This table is how the fictional program keeps training , observation and measurement one system instead of three documents.
Back to top
Content licensed CC BY 4.0 ; code licensed MIT . Bridge Math Tutoring is a fictional organization; all transcripts are fictional and constructed for training purposes.