STEM REASONING AND EVIDENCE WORKBENCH
Summarize anonymous item-level correct and target-option rates with Wilson intervals and minimum-cohort review.
Review output
Load the worked example to inspect realistic inputs, or enter anonymous classroom or laboratory evidence. Results remain in this browser tab until you copy or download them.
Transparent method
Correct rate equals correct divided by attempts. Target rate equals named-option selections divided by attempts. A two-sided 95 percent Wilson interval describes binomial sampling uncertainty. A review signal requires both a cohort at or above the declared minimum and an observed target rate at or above its entered threshold. A threshold signal means the item deserves human review and more evidence. It can reflect conceptual reasoning, wording, instruction, language access, response mapping, guessing, or data error. Inspect explanations and another task before planning a response, and discuss the pattern at cohort level without attaching it to learner identity.
Confirm the question, approved source, units, notation, signs, response coding, model assumptions, and review threshold. Keep an unchanged copy of measurements or learner evidence.
Run the worked example, inspect every intermediate row, and compare the result with a hand calculation, diagram, representation, periodic table, or approved laboratory method.
Record the method, assumptions, uncertainty, units, classifications, and flagged rows. Verify the editable report independently before instruction, assessment, or laboratory use.
How to interpret the result
Use to prioritize follow-up questions or lesson review from aggregate response patterns. It does not diagnose an individual, establish why an option was selected, score open reasoning, adjust for guessing, validate item quality, or justify high-stakes action. Small cohorts are explicitly withheld from threshold decisions.
The fraction item has fourteen of thirty selections on the named option and crosses its thirty-five-percent threshold. Two larger items remain below their thresholds. A six-response pilot shows a high proportion but falls below the declared cohort minimum, so the tool reports insufficient aggregate size instead of treating it as evidence.
Correct rate equals correct divided by attempts. Target rate equals named-option selections divided by attempts. A two-sided 95 percent Wilson interval describes binomial sampling uncertainty. A review signal requires both a cohort at or above the declared minimum and an observed target rate at or above its entered threshold.
A threshold signal means the item deserves human review and more evidence. It can reflect conceptual reasoning, wording, instruction, language access, response mapping, guessing, or data error. Inspect explanations and another task before planning a response, and discuss the pattern at cohort level without attaching it to learner identity.
Do not call every wrong response a misconception, infer cause from a distractor count, expose small cells, compare groups with different item conditions, ignore missing attempts, or treat an interval as a guarantee. Review item alignment, response coding, accessibility, privacy policy, and qualitative learner reasoning.
Learning boundary
This browser-local workbench applies only the disclosed rules. It cannot diagnose a learner, certify an experiment, establish a chemical structure, or replace teacher, laboratory, and safety judgment.
Review checklist
Practical questions
No. Processing occurs in the current browser tab. Avoid personal information and retain authoritative records only in the school-approved system.
No. It describes entered evidence under one disclosed rule. Compare several evidence sources and ask for reasoning before drawing a conclusion.
Treat it as an editable calculation aid. Show your own reasoning, cite required sources, and follow the course method and academic-integrity rules.
YOUR NIRMION
