How score-report analysis works
The purpose of this method is to convert a score report into a defensible study decision. It starts with question-level evidence, measures each pattern with its uncertainty, then maps clear weaknesses to practice.
- 01
Extract the report structure
The parser identifies reported sections, questions, selected answers, difficulty, timing and available metadata. It distinguishes scored sections from unscored or diagnostic coverage.
- 02
Classify each question
Each scored question is filed by question type, LawHub difficulty level and the reasoning patterns behind it. Every type and pattern has a plain-English name and a one-line meaning.
- 03
Measure with honest uncertainty
Accuracy in every group is shown with its question count and a 95% Wilson score range for the true accuracy. A group is called a weakness only when at least 5 questions back it and its whole range sits below the student’s own overall accuracy.
- 04
Rank by points
Clear weaknesses are ranked by the points they cost: the questions the student would have answered correctly in that group at their own average. Timing is reported separately and never inferred from a wrong answer.
- 05
Map the pattern to practice
Each priority maps to an untimed explanation review, a timed set and a mixed transfer set. Drill citations use PTs 139 and lower.
Privacy handling
Uploaded PDFs stay on the server only until the analysis finishes and are then deleted; if an analysis fails, the file is deleted within 24 hours. The saved result holds test and question data, not the student's name or LSAC account number. Lead records are created only when a student submits an email to unlock the detailed report; booking a call creates a booking record. Public research must remove identifying details before aggregation and must state the method and sample limitations. Full details are in the privacy policy.
Limitations
- A score report is evidence about one performance on one test; it is not a complete ability measure.
- One PrepTest has only a handful of questions of each kind, so most ranges are wide. A pattern that repeats on a second report is much stronger evidence.
- Some reports provide more timing detail than others. Missing timing data is stated, not invented.
- Unscored or experimental coverage is left out of every figure.
- The analysis does not claim LSAC affiliation or official status, and it does not predict or guarantee a score.
- Aggregate research must state sample size, date range, and inclusion rules before it is published.