Blind Review and Error Log Method
Key Takeaways
- Blind review separates reasoning ability from timing pressure by making you re-solve uncertain questions before checking answers.
- A useful LR error log records the task, argument core, chosen answer reason, credited answer reason, and repeatable fix.
- Misses should be classified by process failure, such as wrong stem mode, missed conclusion, reversed conditional, or untested assumption.
- Review is complete only when you can explain why the credited answer performs the stem task and why the tempting answer does not.
Why Blind Review Works
Timed Logical Reasoning practice gives two signals at once: what you understand and what you can execute in 35 minutes. Blind review separates those signals. Before checking the answer key, you revisit selected questions without the clock and decide whether your original answer still holds.
This matters because an LR miss can come from different causes. You may not know the underlying reasoning. You may know it but rush the stem. You may identify the gap but choose a true statement that does not answer the question. LSAC's suggested approach warns against selecting a response merely because it is true, and blind review exposes that exact error.
Blind review should not become an excuse to redo the whole section casually. The goal is to recover your best reasoning while the stimulus is still fresh enough to diagnose the timed decision.
What To Blind Review
Review every question you flagged, every question you guessed, and every question where confidence was low even if you selected an answer.
Do not check whether you were right before blind review. If you know the answer, the work changes. You start justifying instead of solving. The question is: with more time, what would I choose and why?
Blind Review Routine
| Step | Action | Output |
|---|---|---|
| 1 | Hide the answer key | No confirmation bias |
| 2 | Re-read the stem | Exact task label |
| 3 | Rebuild the stimulus | Premise, conclusion, gap, or fact map |
| 4 | Re-evaluate all choices | One reason to keep or eliminate each |
| 5 | Lock a blind-review answer | Timed answer versus best-reasoned answer |
| 6 | Check the key and explanation | Error category and fix |
The locked blind-review answer is important. If your timed answer was wrong but blind review is right, the main issue is execution under time. If both are wrong, the issue is understanding or answer evaluation. If timed is right but blind review changes away from it, the issue is confidence and overthinking.
Build The Error Log
An error log is not a list of wrong questions. It is a list of repeatable decisions. Use columns that force analysis.
Recommended columns: question source, task, timed answer, blind-review answer, credited answer, confidence, time used, argument core, error type, and next rule. The argument core should be one sentence, not a paragraph. For example: because survey respondents preferred the online form, author concludes all applicants will use it, assuming respondents represent applicants.
Error Categories
| Category | What happened | Fix rule |
|---|---|---|
| Stem mode | Treated inference like weaken, or flaw like weaken | Say the stem verb before choices |
| Core map | Misidentified conclusion, premise, speaker, or rule | Use because/therefore or speaker labels |
| Gap | Saw topic but not missing link | Compare premise terms to conclusion terms |
| Scope | Changed group, time, quantity, or standard | Circle quantifier and target group in review |
| Formal | Reversed conditional or missed contrapositive | Translate only-if/unless before choices |
| Answer trap | Picked true, familiar, or extreme claim without task fit | Require a job-specific reason |
| Timing | Stayed after progress stopped | Set skip trigger for that task family |
The fix rule must be behavioral. "Be more careful" is not a fix. "On necessary-assumption questions, negate finalists only after writing the gap" is a fix.
Explain Both Sides
Review is not finished when you know why the credited answer is good. You must know why your tempting answer failed. Wrong answers in LR often sit close to the argument. They may use the same topic, describe a real concern, or make a statement that is factually plausible. The test asks whether the answer performs the requested job.
For each miss, write two sentences. First: the credited answer is right because it does the task by doing X. Second: my answer is wrong because it fails by doing Y. Keep the language structural: alternate cause, reversed rule, unsupported scope, one-speaker-only, true but irrelevant, proves nearby claim.
This two-sentence method prevents passive rereading of explanations. It also prepares you for future questions that use different subject matter but the same trap.
Review Correct Answers Too
Correct answers deserve lighter review, but not zero review. Mark any correct answer that took too long, any correct answer you changed late, and any correct answer with low confidence. A correct answer chosen for a bad reason is unstable.
For high-confidence correct answers, look for process worth repeating. Did you prephrase the causal gap? Did you translate an only-if rule cleanly? Did you reject a true statement because it did not answer the stem? Add these as positive rules.
Weekly Pattern Work
At the end of each week, count error categories rather than total wrong answers. If six misses were stem-mode errors, do not drill more formal logic yet. If four misses were overlong parallel items, practice abstraction and earlier flagging. If many misses were blind-review-correct, train timed sets with stricter checkpoints.
Use a pattern summary like this:
| Pattern | Evidence | Next drill |
|---|---|---|
| Necessary assumption overstrong | Three misses chose sufficient support | Ten-question NA set, negate modest finalists |
| Causal weaken timing | Two correct but over two minutes | Five causal weakens, prephrase alternate cause |
| Parallel fatigue | Accuracy drops after question 15 | Two back-to-back sections with early heavy flags |
Blind review is successful when your log changes your next timed behavior. The section score is a snapshot. The review is the training.
After a timed LR set, a student checks the answer key immediately, sees a missed weaken question, reads the explanation, and writes "careless" in a notebook. What would most improve the review process?