Error Pattern Tracker

#WST01 — Error Pattern Tracker (Statistics 1)

Provenance (2026-09-29). Built from research/veridian/WST01-verified-facts.md (compiled 2026-09-12, spot-check addendum 2026-09-25) — specifically §5 (per-topic verified content), §6 (the cross-cutting "show that" pattern), §7 (topics with enough verified material for a standalone lesson) and §9 (the 2026-09-25 gap-closing audit of June 2019). Cross-checked against forge-exam-technique-guide.md §5 (its own trap-index table, so nothing below silently restates it) and signal-frequency-and-boundaries.md (the 6-series methodology used for every "N of 6" count below: Jun 2019, Jan 2021, Oct 2021, Jun 2022, Jan 2023, Jun 2024 — the only 6 of the 14 verified series with a fully-read Principal Examiner Feedback report). Series counts below were independently re-tallied from the facts bank's own citations for this document, not copied from its prose — one correction came out of that re-tally, flagged in §1, item 7. No new PDFs were fetched for this pass. Not affiliated with or endorsed by Pearson Edexcel. Short quotes under ~15 words carry their original attribution; everything else is paraphrase. No likelihood percentage or grade boundary appears anywhere below — every frequency is stated as "seen in N of 6 series checked," never a predicted probability, and never invented for a series that wasn't actually checked.

#What this is

The facts bank and the FORGE guide already built WST01's taxonomy — every individual trap, named, cited to a real series and question, and pointed at the lesson file that teaches it in full (forge-exam-technique-guide.md §5.1's own index). This document does not repeat that list. It is the missing piece: a tool for logging a student's own timed-practice mistakes against that taxonomy, organised the way a student actually needs it mid-self-mark — first by the handful of structural patterns worth checking on every paper, then by spec section for the narrower, topic-specific traps. Re-type or photocopy the blank tables in §3 and §5 before each timed sitting; fill them in during the self-mark that follows the paper, not during the paper itself.

---

#1. Seven recurring patterns, pre-populated (not a student's to invent)

These are the mechanisms the facts bank found repeated across multiple series and multiple questions, rather than the ~45 individual, single-citation traps §4 indexes one at a time. A timed paper's real mistakes should mostly sort into these seven; if one genuinely doesn't fit, log it under "other" in §5 rather than forcing it into a box that doesn't apply.

Correction made compiling this table. The facts bank's own §5 prose describes item 7 as "confirmed independently in 3 separate series," but lists only two distinct sittings (Jan 2021, and Jun 2024 twice, for two different questions) — re-tallying the actual citations for this document found 2 series, not 3. The mechanism is still real and still worth its own row; the count above is the corrected one, not the facts bank's original phrasing.

Patterns 2, 3, 4 and 6 are, by this recount, the highest-frequency of the seven (4 of 6 series each) — worth checking first when a paper comes back with marks missing and no obvious reason why.

---

#2. How to self-mark in a way this tracker can actually use

Do this immediately after finishing a timed paper, before consulting the mark scheme in detail:

1. For every part that lost marks, write in the margin which of the seven patterns above it looks like, in one or two words. A rough guess is enough to start — you're checking it next.
2. Open the mark scheme (or, for the paper's own real worked anchors, content/veridian/ question-bank/wst01.ts's two full marked-solution entries — June 2019 Q2(c) and Q5(e)) and check the guess against what was actually credited, M/A/B line by line.
3. Log it in §3 (one of the seven structural patterns) or §5 (a narrower, topic-specific trap — check §4's per-spec-section index first for the exact name before assuming it's new).
4. On a Normal-distribution question specifically, check whether a sketched curve with the relevant region shaded would have caught the error before you committed to an answer — Pearson's own examiner reports recommend this twice, verbatim, for exactly the errors in patterns 2 and 3 (Jan 2023 Q5(a): "a simple diagram would have helped many to avoid this error"; Jan 2021 Q3(b): "an error that could probably have been avoided if they had drawn a suitable diagram").
5. After three or more papers, total the tally columns in §6. A pattern with three or more entries in a student's own log is a stronger, more personal signal than any of the frequency counts in §1 — it's this student's actual weak spot, not the cohort's.

---

#3. Structural-pattern log (the seven from §1)

One row per mistake, one table per practice paper. Leave "Fix applied next time?" blank until the next paper, then go back and tick it once that fix has genuinely been tried.

---

#4. Topic-specific trap index, by spec section (for §5's lookup — not a restatement)

Every row below already has a full worked treatment in the lesson named. This index exists so a mistake can be matched to its real name during a self-mark without re-reading the whole FORGE guide. Two rows are marked NEW — they were added to the curriculum on 2026-09-25 (facts bank §9) and are not yet present in forge-exam-technique-guide.md §5.1's own pointer table; this tracker is, as of this pass, the first WST01 technique-hub document to index them by name.

Section 2 — Representation and summary of data (2.1–2.4)

• histogram-height-read-as-frequency — Jan 2023 Q1(a) → reading-and-comparing-distributions.ts
• stem-and-leaf-read-in-the-wrong-direction — Jun 2024 Q1(b) → reading-and-comparing-distributions.ts
• mean-denominator-is-class-count-not-total-frequency — Jan 2023 Q1(c) → measures-of-location-and-dispersion.ts
• coding-decode-forgot-to-square-constant / coding-decode-multiplied-instead-of-divided — Jun 2022 Q3(e) → measures-of-location-and-dispersion.ts
• sd-formula-missing-divide-by-n / sd-formula-square-root-in-wrong-place — Jan 2023 Q1(d), Jan 2021 Q6(c) → measures-of-location-and-dispersion.ts
• n-and-n-plus-1-treated-as-identical — Jan 2023 Q1(e) → measures-of-location-and-dispersion.ts
• outlier-lower-fence-direction-reversed, whisker-drawn-to-fence-not-to-data — Jan 2021 Q2 → outliers-boxplots-comparing-distributions.ts
• remembered-formula-overrides-the-rule-given-in-the-question — Jun 2022 Q1(b) → outliers-boxplots-comparing-distributions.ts
• NEW amended-value-fences-not-recalculated — Jun 2019 Q2(c): a corrected data value shifted one quartile, forcing both fences to be rebuilt even on the side that didn't itself change → outliers-boxplots-comparing-distributions.ts
• mean-assumed-more-accurate-than-the-median, comparison-missing-supporting-figures — Jun 2022 Q1(e), Jun 2024 Q1(e), Oct 2021 Q3(e) → outliers-boxplots-comparing-distributions.ts, reading-and-comparing-distributions.ts

Section 3 — Probability (3.1–3.4)

• conditional-probability-as-raw-ratio-of-marginals — Jan 2021 Q1(c) → elementary-probability-and-conditional-notation.ts, conditional-probability-independence-vs-mutually-exclusive.ts
• distinct-events-assumed-mutually-exclusive / mutually-exclusive-and-independent-treated-as-compatible — Oct 2021 Q1(b) → conditional-probability-independence-vs-mutually-exclusive.ts
• independence-assumed-instead-of-tested — Jun 2022 Q4(b), general comment → sampling-tree-diagrams-venn-diagrams.ts, conditional-probability-independence-vs-mutually-exclusive.ts
• original-fraction-reused-after-removal, denominator-kept-constant-across-repeated-draws — Jan 2023 Q2, Jun 2022 Q3(d) → sampling-tree-diagrams-venn-diagrams.ts
• venn-blank-region-assumed-zero — Jun 2022 Q4(c) → sampling-tree-diagrams-venn-diagrams.ts

Section 4 — Correlation and regression (4.1–4.3)

• sxx-computed-with-xbar-squared (real evidence is for $S_{pp}$, structurally identical) — Oct 2021 Q2(b) → regression-gradient-interpretation-extrapolation.ts
• fraction-left-in-final-line — Jun 2022 Q2(c) → regression-gradient-interpretation-extrapolation.ts
• gradient-interpreted-at-the-wrong-scale, units-of-y-dropped-or-swapped — Oct 2021 Q2(d), Jun 2022 Q2(d), Jan 2023 Q6(a), Jun 2024 Q4(d) → regression-gradient-interpretation-extrapolation.ts
• range-boundary-not-checked — Jun 2022 Q2(e), Jan 2023 Q6(c) → regression-gradient-interpretation-extrapolation.ts
• reverse-gradient-solve-not-attempted — Jun 2022 Q2(f) → regression-gradient-interpretation-extrapolation.ts
• sqrt-omitted-in-pmcc-denominator — Jun 2024 Q4(b)/(c) → correlation-coefficient-and-regression-calculation.ts
• coding-invariance-not-recognised (r unaffected by linear coding) — Jan 2024 mark scheme Q2(d) → correlation-coefficient-and-regression-calculation.ts

Section 5 — Discrete random variables (5.1–5.4)

• sum-to-1-check-skipped, domain-reinterpreted-to-avoid-the-gap — Oct 2021 Q4(d), Jun 2022 Q5(d) → discrete-random-variables-probability-function.ts
• coefficient-not-squared-in-variance — Jan 2021 Q4, Jun 2024 Q2(c)/Q3(d) → expectation-and-variance-of-linear-transformations.ts
• NEW positive-expectation-mistaken-for-a-favourable-single-play-probability — Jun 2019 Q5(e): E(X) > 0 was cited as the reason to play a game whose actual win probability was below 0.5 → discrete-random-variables-probability-function.ts

Section 6 — The Normal distribution (6.1)

• subtracted-from-1-when-not-required (and its mirror, not subtracting when it was required) — Jan 2021 Q3(a), Jan 2023 Q5(a), Jun 2024 Q5(a) → normal-distribution-standardisation-and-conditional-probability.ts
• rounded-z-value-instead-of-4dp-table-value — Jan 2021 Q3(b), Oct 2021 Q6(b), Jun 2024 Q5(b) → normal-distribution-standardisation-and-conditional-probability.ts
• sign-error-reversing-standardisation — Jan 2021 Q3(b), Oct 2021 Q6(b) → normal-distribution-standardisation-and-conditional-probability.ts
• conditional-probability-not-recognised-with-normal — Jan 2021 Q3(c), Oct 2021 Q6(c), Jan 2023 Q5(e) → normal-distribution-standardisation-and-conditional-probability.ts
• show-that-standardisation-not-shown — Oct 2021 Q6(a) → normal-distribution-standardisation-and-conditional-probability.ts, show-that-answer-discipline.ts
• Genuinely unnamed gap, disclosed rather than forced into an existing box: Jun 2024 Q5(c), a multi-step inverse-Normal problem, drew this from the examiner report: "it was not unusual for students to give up at this point" — real, quoted, verified evidence of candidates abandoning a problem partway through rather than making a specific calculation error. No misconception id in normal-distribution-standardisation-and-conditional-probability.ts currently names this "abandons a multi-step problem" pattern specifically (the file's ids all describe a wrong calculation, not a stopped one). Log an instance of this under "other" in §5 until the curriculum names it.

---

#5. Topic-specific trap log

For a mistake that matches a named trap in §4, write the trap's own name rather than a fresh description — the name links straight back to the lesson file that drills it, so a repeated entry here is an instruction to re-open a specific page, not just a record that something went wrong.

If a mistake doesn't match any named trap in §4 and doesn't fit one of the seven patterns in §3 either, log it here as "other." A repeated "other" entry across two or more papers is real, disclosed new evidence — exactly the finding that should be raised for the facts bank's own §8 or a future audit pass to absorb, not quietly dropped because it doesn't fit an existing box.

---

#6. Running tally (fill in after every third paper)

Read this table across, not down, after the fifth paper. A pattern present in four or five of the five columns is not one bad paper — it is this student's single most valuable thing to fix before the next attempt, ahead of anything else in this document.

---

#7. Two worked example entries (real, sourced, filled in to show the format)

These are not a student's own mistakes — they are the paper's own confirmed error instances, entered once each to show how a completed row should look.

§3 example (pattern 2, rounded value instead of exact/4-d.p.):

§5 example (named trap):

Both rows draw on real, already-cited examiner-report evidence (§1, §4) — nothing in either row is invented for this demonstration.

---

#8. What this document deliberately leaves out

• No log rows for spec section 1.1 (mathematical models). The facts bank confirms no standalone question testing this exists anywhere in the 14-series archive it checked — content here is examined embedded inside other questions (e.g. "assume the die is fair"), not as its own part. A tracker category needs a real, citable error instance to populate it; none exists for this spec point, so none is invented here.
• No grade boundaries and no predicted-likelihood figures. Neither exists in any source checked for this paper (signal-frequency-and-boundaries.md §2 records the grade-boundary gap specifically). Every frequency above is "seen in N of 6 series checked," not a probability.
• 8 of the 14 verified series still have no located examiner report (Jun 2021, Jan 2022, Oct 2022, Jun 2023, Oct 2023, Jan 2024, Oct 2024, Jan 2025) — the facts bank's own §8/§9 treat the current trap taxonomy as a lower bound, not a finished inventory, and this tracker inherits that limit. A pattern that isn't in §1 or §4 yet may simply not have been surfaced by a fully-read report yet — that is why §5's "other" table exists and why it explicitly asks a student to log a repeated unmatched mistake rather than discard it.
• This document does not re-print the ~45 individual trap descriptions the FORGE guide already indexes in full at its own §5.1 — duplicating them here would drift out of sync with that index the first time either document is next updated. §4 gives the name and citation only; look the full worked treatment up in the lesson file named.