Unofficial guide · not affiliated with the NCBE, the ABA, or any state bar ← SI Bar Exam · Exam Admin · Academy

Eli Item-Writing Academy

Academy › Quality at Scale

Peer Review: The Cold-Solve Protocol

About 30 minutes · Academy module: Peer Review: The Cold-Solve Protocol.

Learning goals

  • Run a two-solver cold review on every item before it ships.
  • Give actionable item feedback: tested point, key defensibility, distractor quality.
  • Log review outcomes so disputes are traceable.

Lesson 1 — Why Cold Solvers

The author cannot review their own item — they know the tested point, so every ambiguity is invisible to them. Cold solvers (colleagues who haven't seen the item) replicate the examinee's position. The protocol: solver reads the item once, selects an answer, and writes one sentence on why each distractor fails. No discussion before both solvers commit.

Lesson 2 — The Review Form

Four judgments per item: (1) Tested point — can the solver state it in one sentence? If not, the item is muddy. (2) Key — did both solvers converge? If not, diagnose (see the Single-Best-Answer module's two-key rescue). (3) Distractors — did each attract at least consideration, and can the solver name each one's error? (4) Sensitivity — any gratuitous identity detail or loaded scenario?

Lesson 3 — Disagreement Is Data

When solvers disagree, the item is telling you something. Common diagnoses: ambiguous facts (two readings), a distractor that's defensible on the call as written (sharpen the call), or a tested point the author never stated (write it down, rebuild). Log the disagreement and the fix — the log is what makes your bank's quality auditable.

Lesson 4 — Review At Scale

For large banks, sample: every new item gets two solvers; a quarterly 10% re-sample catches drift. Track solver agreement rates per author — not to punish, but to target coaching. The Question Studio's built-in two-checker gate automates this protocol for items drafted with Eli.

Spot-the-flaw drill

Flaw: no independent review — the author's blind spots ship with the items. Within a w

no independent review — the author's blind spots ship with the items. Within a week, students report three disputed keys; the author spends more time firefighting than the review would have cost.

Correction

The minimum viable protocol: one colleague, cold, 20 items, one sitting. Even a single cold solver catches most key disputes and unfocused stems.

Model rewrite

— the review log entry:

Item CIV-S07 (Erie): Solver A chose key (B), Solver B chose (D). Diagnosis: fact pattern ambiguous about whether the federal rule was "arguably procedural" — B and D defensible on different readings. Fix: added one sentence clarifying the rule's function; both solvers re-solved cold → B, B. Shipped.

Drill key

The log shows the disagreement, the diagnosis, the fix, and the re-solve. That paper trail is the quality system.

Sources

  • Downing, S. M., & Haladyna, T. M. (Eds.). (2006). Handbook of test development. Lawrence Erlbaum Associates. (On item review panels and reviewer training.)

Finished this module?

Recording it adds the module to your Academy completion record on this device — module, track, level, and date, ready to download from your account page for faculty-development documentation.