A percentage without an error log is a feeling with a number. Categorize misses, match each type to a repair, and let mocks drive next week — not just mood.
You finish a mock. The score lands. Then one of two things usually happens: celebration that evaporates by Thursday, or a shame spiral that produces another untimed paper binge.
Neither is feedback.
John Hattie and Helen Timperley’s model of effective feedback asks three questions: Where am I going? How am I going? Where to next? A percentage answers a thin version of the second question and ignores the third. Barry Zimmerman’s self-regulated learning cycle — plan, perform, reflect — collapses at reflect if reflection is only “72%… okay.”
An error log is how you turn markschemes into fuel. Different error types need different fixes. Treating every miss as “study more” is how weekends disappear.
The four-type taxonomy (keep it light enough to use)
You do not need a research lab. You need one line per miss.
| Error type | What it means | Next session |
|---|---|---|
| Misconception | Wrong or incomplete mental model | Re-encode: guide, generative notes, tutor scaffolds, teach-back |
| Slip | Knew it, fumbled execution or wording | Targeted retrieval; honest ratings; similar item later |
| Time fault | Ran out / rushed / poor allocation | Timed sections; pacing strategy; triage rules |
| Gap | Never learned / missing prerequisite | Full encode loop before more papers |
If you are unsure between slip and misconception, try a closed-book teach-back of that concept. If the model collapses, it was not a slip.
Why categories beat vibes
Self-regulation requires control decisions — what to do next. Categories map to methods. Vibes map to guilt. Guilt is not a study plan.
How to log without turning review into a second exam
Right after marking (while memory of your thinking is fresh):
- For each lost mark cluster, write: concept + type + one-sentence cause
- Star anything that is high-weight on the syllabus
- Cap the repair queue at 3–5 items for the week
- Assign a method to each item before you leave the desk
- Schedule the methods on the calendar (encode / retrieve / timed section)
Example log lines:
- “Enzyme inhibition — misconception — mixed up competitive vs non-competitive graphs”
- “Q4 algebra — slip — dropped a negative under time”
- “Section B — time fault — wrote essays in order instead of marks-per-minute”
- “Electrolysis prep — gap — never learned half-equations cleanly”
That is enough. Novels are not required.
Match repair to type (the step everyone skips)
Misconception: Stop doing more papers on that topic until you can teach the corrected model. Clarity and generative encoding first; then a probe; then a similar question.
Slip: Do not rewrite the whole chapter. Retrieve the cue under slight delay; practice the exact failure point; rate honestly.
Time fault: Practice allocation. Timed mini-sections beat another leisurely perfect paper. Decide skip/return rules in advance.
Gap: Prerequisite repair. Sweller’s cognitive load theory predicts pain if you keep sitting papers that assume schemas you never built.
The ‘rewrite the whole paper’ trap
Copying a full perfect script after marking feels responsible and often creates markscheme fluency without model repair. Reconstruct reasoning for misconceptions; don’t transcribe the key.
From log to next week’s plan
Your weekly plan should be readable from the log:
- Mon: encode session for top misconception
- Tue: retrieval block including slip items
- Wed: timed section practicing pacing
- Thu: mixed questions on starred concepts
- Sun: re-check teach-backs; update focus list
This is feed-forward feedback in calendar form. Dashboards and focus lists help only if the log supplies honest inputs.
How often to revisit the log
- After every full paper: add lines
- Mid-week: confirm repairs were scheduled (adherence)
- Before the next paper: skim open items — do not sit another full paper while the same misconception is untouched
If the same misconception appears twice, change method. More volume of the same broken loop is not grit; it is repetition of an error.
Papers without logs are expensive entertainment
Full papers are high-value because they create transfer-appropriate practice and a natural interleaved mix. That value evaporates if the only artifact is a score. The log is the artifact that makes the sitting worth the emotional cost.
Deliberate practice needs a diagnosis, not a vibe
Anders Ericsson’s deliberate-practice framework requires targeted work on specific weaknesses with feedback — not more hours of the same. An error log is how a mock becomes deliberate. Without types, another paper is only more performance. With types, Monday’s encode session is aimed at the exact model break the markscheme exposed.
Metcalfe and others’ work on error correction also warns that uncorrected errors can persist; logging without scheduling the repair leaves the wrong model in place while you collect new scores. The log line is a promise to the calendar.
Feedback that feeds forward
Hattie and Timperley’s ‘where to next’ is empty without a method. ‘Study enzymes’ is not feed-forward. ‘Re-encode competitive vs non-competitive graphs; teach-back; two probes Thursday’ is.
Almost-right answers deserve their own star
Partial credit items are often richer than blank misses. You had a model — a wrong branch or a missing boundary. Log the branch. “I treated this as equilibrium when it was kinetics” is gold. “I got 4/6” without a cause is fog. Almost-right is where misconceptions hide behind fluency.
Keep the taxonomy stable across the term
Switching category systems every week kills adherence. Four types are enough: misconception, slip, time fault, gap. If you invent twelve micro-labels, the log becomes a second exam. Consistency lets patterns appear — the same misconception across papers is a method problem, not a motivation problem.
Review open items before you sit the next full paper. Sitting again with the same untouched misconception is expensive confirmation bias: you already know what will break.
Teachers and shared error language
When teachers and students share the same light taxonomy, feedback conversations get shorter and more actionable. “Time fault on Section B” points to pacing drills. “Misconception on titration curves” points to re-encode. Shared language turns mark comments into next actions instead of vague “revise topic 4.”
Close the loop with a re-check
A log entry is not done when the repair session ends. Schedule a production re-check: teach-back, similar item, or timed mini-section. Only then archive or demote the line. Without re-check, logs become museums of good intentions. With re-check, they become evidence that control worked — Zimmerman’s cycle completed.
Sit, score, then steer
Sukrat Mock Exams support full-paper practice with review that should feed what you fix next — pair scores with a simple error log and a planner block. Product page: /products/features/mock-exams
FAQ
Should I log every single mark?
Log clusters and concepts. Obsessive micro-logging dies by week two.
What about almost-right answers?
Often the richest misconceptions. Log them even if you “mostly got it.”
Can AI write my error log?
It can help classify after you describe your thinking. If the AI invents causes you did not experience, you lose the metacognitive training.
Related reading: Mock Exams cluster
- Full Papers Train Transfer
- Past Papers Without Memorizing Markschemes
- From Mock Score to Next Week’s Plan
- Timed Practice Builds Exam Stamina
- Exam Transfer vs Recognition
- How Often to Sit Full Papers
Keep going across Sukrat
Scores tell; logs steer
Mocks are diagnostic instruments. Without an error taxonomy, they become mood machines — up after a high score, useless after a low one.
Categorize. Match a method. Schedule the repair. Re-check with production. That loop is self-regulated learning with evidence.
A percentage is information. A log is a plan.