LSAT score-report task-accuracy dataset
This page reports aggregate patterns from score reports processed by the independent LSAT Lawyers diagnostic. It is intended to make the sample, method, privacy handling, and limits explicit before the data is interpreted.
Original dataset · Published 2026-08-28 · Updated 2026-08-28
- Unique reports
- 17
- Unique tests
- 14
- Duplicate files removed
- 22
- Median score
- 160
- Score range
- 148–180
Direct interpretation
The current aggregate contains 17 deduplicated reports from 14 distinct tests; 39 valid score-report files were scanned and 22 duplicate copies removed. The median score is 160 and the range is 148–180. This is a small, non-random sample, so it must not be generalized to all LSAT test-takers. It shows task-level separation (Inference, Paragraph Purpose, Must Be True, Parallel Flaw, and Application rank lowest) rather than a universal difficulty ranking.
Task accuracy table
Ranked across task families with at least 15 attempted questions in the deduplicated sample.
| Task family | Attempts | Misses | Accuracy | Reports |
|---|---|---|---|---|
| Inference | 75 | 33 | 56% | 17 |
| Paragraph Purpose | 23 | 10 | 56.5% | 12 |
| Must be True | 46 | 19 | 58.7% | 16 |
| Parallel Flaw | 44 | 18 | 59.1% | 17 |
| Application | 24 | 9 | 62.5% | 15 |
| Author Opinion | 73 | 27 | 63% | 17 |
| Primary Purpose | 37 | 13 | 64.9% | 14 |
| Method | 46 | 15 | 67.4% | 14 |
| Weaken | 103 | 32 | 68.9% | 17 |
| Local Purpose | 39 | 12 | 69.2% | 16 |
Method
- Locally retained score-report PDFs that passed the independent parser and structural preflight.
- Exact duplicate report structures removed before aggregation.
- Experimental sections excluded by the analyzer.
- Deduplication: Exact report-structure fingerprint before aggregation
- Only task families with at least 15 total attempted questions across unique reports are ranked.
Privacy
Public output contains aggregate task accuracy and counts only; no names, emails, IPs, job IDs, local file paths, selected answers, raw responses, or individual PrepTest identifiers are published.
Public aggregate mirror
The aggregate-only JSON is mirrored on GitHub and as a standards-compliant machine package for verification and reuse. The canonical research page remains this URL.
Terms
Cite the canonical URL and date. Aggregate figures may be quoted with the sample size. Do not present this as LSAC data, official LSAT data, population-level research, or a guarantee about tutoring results.
Limitations
- This is a small, non-random sample of retained score reports and submissions to one independent tool.
- The aggregate includes reports from tutoring and testing workflows, so it is not a random sample of all LSAT students.
- Accuracy rates describe these unique reports only and should not be generalized to all LSAT students.
- Exact-structure deduplication removes duplicate copies, but repeated independent submissions of the same performance cannot be excluded completely.
- Task labels depend on the independently maintained question metadata used by the diagnostic.
- No causal, tutoring-effectiveness, or score-outcome claim is made.
Find your pattern
Compare your score report against these task-level patterns instead of guessing from a section score.