Reasoning diagnostics for AP Physics, F=ma & USAPhO mechanics

Find the physics model behind each miss.

Practice sites tell you what you missed. This maps why.

FirstPrinciples maps each diagnostic choice to a named reasoning trap, then routes you to the Atlas entry and ladder rung written to repair it.

one miss, traced
  1. answerQ04 · B
  2. trapcircular-direction
  3. atlasNL-05
  4. rungR7
try it below ↓

Live example: one diagnostic item, routed

Diagnostic bench item Q04 · mechanics · circular motion · AP Idle

A car rounds a flat curve at constant speed. The direction of the net force on the car is:

Pick any answer, right or wrong. Nothing is scored or saved.

signal awaiting an answer▊
  1. 01
    Answer —
  2. 02
    Reasoning pattern the model behind the choice
  3. 03
    Misconception code one of 12 trap families
  4. 04
    Explanation · Error Atlas the entry that explains the trap
  5. 05
    Repair · ladder rung the problem that rebuilds the idea

The full diagnostic runs this on 48 items, then ranks the traps that recur.

Run the 48-item diagnostic →

01 The correction loop

Every miss runs through the same six steps.

The same coding rules power the report, the Atlas links, and the recommendations — shown here for the miss above.

The correction loop: answer pattern, misconception code, diagnostic report, Error Atlas entry, recommended rung, progress data — then repeat attempts show which traps persist.
  1. 01
    Answer patternWhich choice, on which item.Q04 · chose B
  2. 02
    Misconception codeEvery distractor carries a trap code.circular-direction
  3. 03
    Diagnostic reportCodes are tallied; traps that recur rank first.ranked by recurrence
  4. 04
    Error Atlas entryWhy the wrong model felt right, and what replaces it.NL-05
  5. 05
    Recommended rungOne problem that adds the missing idea.R7
  6. 06
    Progress dataA held-out follow-up checks whether the trap recurs.recurring / resolved

02 Three surfaces

Diagnosis, atlas, ladder — wired together.

Practice alone rehearses mistakes. Here, each attempt decides what you read next and which problem you work next.

03 Verified reach

Where the numbers come from.

65 of 19,185 reached the verified cohort; 45 have matched pre/post pairs. Each figure is drawn to scale, with the claim it can support.

Full methodology →
  1. 19,185prep users

    Broad reach. Supports reach. It does not prove learning gain.

  2. 2,102diagnostics completed

    Platform adoption. Supports product usage and diagnostic adoption.

  3. 65verified cohort

    Research sample. Supports cohort-scale analysis.

  4. 45paired pre/post

    Learning evidence. The layer used for learning-impact claims. Pre/post evidence, not a controlled trial.

04 The open question

Does knowing which reasoning step broke actually help more than doing another practice set?

Baseline diagnostic, targeted ladder work, held-out follow-up, paired comparison. Results get published with the denominator attached.