Exam Technique Guide

#WST01 (Statistics 1) — FORGE Exam-Technique Guide

Provenance: written 2026-09-28, built on research/veridian/WST01-verified-facts.md (compiled 2026-09-12, spot-check addendum 2026-09-25) and cross-checked against the 13 real lessons already live in content/veridian/curriculum/wst01/. No new PDFs were fetched for this document — the facts bank's own §3 records that, as of its 2026-09-25 revision, no local WST01 PDF archive exists anywhere on this machine and every source was re-verified fresh over the network; a fresh search of every working directory available to this session (~/Downloads/PAPERI, ~/Documents/SOL/Papers, ~/Documents/SOL/VERIDIAN, ~/Downloads/exaxm technique, ~/Downloads/IAL_Paper3_Materials, the Master_Consolidated_Docs folders) turned up no WST01/Statistics material either — confirmed, not assumed. This guide is a technique-and-craft layer synthesised from that already-verified record; it does not re-teach the content lessons and does not claim any independent PDF verification of its own.

#How to use this document

This is not a content lesson — it assumes you already know how to standardise a Normal variable, fit a regression line, and build a probability function, because the 13 lessons in content/veridian/curriculum/wst01/ teach those. What this document is: everything about how the exam itself is built and marked that sits above any one topic — the timing, the M/A/B system, the notation conventions the mark scheme actually enforces, and a single cross-referenced digest of every verified error pattern in the record, organised by exam-technique category rather than by spec section. Where the task brief that generated this document assumed a feature of this paper that turns out not to exist (see §3.2 below), that is stated plainly rather than invented around.

---

#1. Paper structure and a real timing budget

The numbers, confirmed on all 14 series' front pages this course's research pass reviewed (Jun 2019 → Jan 2025), and independently re-checked directly against a real paper in paper-anatomy.ts:

• 90 minutes, 75 marks. Fixed on every series checked.

• No Section A/B/C. WST01 is one flat sequence of compulsory questions, unlike the Economics/ Business papers on this qualification. Every question ends in its own printed "(Total for Question N is X marks)" line.

• The number of questions is not fixed — it is real, verified, and easy to get wrong if you assume otherwise:

  Series	Questions
  Jun 2019, Jan 2021, Jun 2021, Oct 2021, Jun 2022, Oct 2023, Jun 2024	6
  Jan 2022, Oct 2022, Jun 2023, Jan 2025	7
  Jan 2024, Oct 2024	8

  Later series trend toward more, shorter questions. Do not walk into the exam expecting 6 questions — check the front page's own "There are N questions in this question paper" line before you do anything else.

• All questions are compulsory — no choice, confirmed on both the spec (§S1.2, "Students must answer all questions") and a real front page ("Answer all questions... Try to answer every question").

• No multiple-choice item exists anywhere in the real record. Every item is free-response, multi-part. (The MCQ blocks inside this course's own WST01 lessons are retrieval-practice tools, not a reproduction of the real exam's format — worth knowing so the format itself is never a surprise.)

Timing. The flat average is 90 minutes ÷ 75 marks = 1.2 minutes (72 seconds) per mark — and Pearson's own front page states the pacing rule directly, in its own words, not this document's arithmetic: "The marks for each question are shown in brackets – use this as a guide as to how much time to spend on each question." The one full real per-question breakdown this course has independently verified (January 2023, six questions) makes clear why the flat average is only a starting point, not a plan:

The spread is the real lesson, not the average: the smallest question on that series is barely a tenth of the paper; the largest — Normal distribution plus conditional probability — is close to a quarter of the entire exam in one question. That imbalance is structural, not a one-series accident: Normal-distribution material is, independently, the richest and most consistently examiner-flagged content in the whole facts bank (see §5 below), and it is reasonable to expect it to carry a disproportionate mark share on most series, not just this one. A practical rule this document adds, grounded in that spread rather than invented: don't ration time evenly across however many questions the front page says there are — read every question's own printed mark allocation first, and expect the Normal-distribution question specifically to run long.

Cross-reference, not duplication: the fuller version of this section — the real command-verb vocabulary, the AO1–AO5 weighting table, and why WST01 carries no AO tags on individual mark-scheme lines — is paper-anatomy.ts, already live in the curriculum. This section restates only the timing numbers a technique guide needs on one page.

---

#2. The M/A/B mark-coding system, as this paper's real mark schemes use it

Verified verbatim from the "General Instructions for Marking" boilerplate that opens every one of the 14 mark schemes reviewed — confirmed byte-for-byte identical (barring the total-marks figure) across at least 6 of the 14 spot-checked directly (Jan 2021, Jan 2023, Jan 2024, Jun 2021, Jan 2022, Oct 2022). This is paper-wide Pearson IAL Maths boilerplate, not bespoke Statistics wording — paper-anatomy.ts even documents a Mechanics-flavoured example sentence ("resolving in a particular direction, taking moments about a point") left verbatim inside a real WST01 mark scheme that never uses moments or suvat equations itself, direct proof this text is shared across the whole qualification suite (P1–P4, M1–M3, S1–S3, D1).

#2.1 The three mark types

• M — method. "A method mark is given for a correct method or an attempt at a correct method... M marks are sometimes dependent (DM) on previous M marks having been earned." An M mark rewards the attempt at the right process, not a correct final number.
• A — accuracy. "Dependent accuracy (or sometimes answer) marks and can only be awarded if the previous M mark has been earned. E.g. M0 A1 is impossible." This single sentence is the whole mechanism behind "show that" discipline (§4 below, and the dedicated lesson show-that-answer-discipline.ts): if the method line isn't visible, every accuracy mark chained to it is unreachable, however correct the final value is.
• B — independent/standalone. "Independent accuracy marks where there is no method (e.g. often given for a comment or for a graph)." A B mark doesn't depend on an M mark at all — it rewards a fact, a comment, or a correctly-placed diagram feature standing on its own.
• ft (follow-through). "A few of the A and B marks may be f.t. — follow through — marks," meaning the mark can still be earned using an earlier (possibly wrong) value, provided the method applied to it is correct.

#2.2 Abbreviations that actually change what you should write

Verbatim, from the mark schemes' own "General Abbreviations" list:

#2.3 The general marking principles — genuinely new to this course's WST01 material

The following are verbatim from the mark schemes' own "General Marking Guidance," verified in WST01-verified-facts.md §4, but not yet present in any of the 13 existing WST01 lessons — a genuine gap this document closes rather than duplicates:

"All candidates must receive the same treatment. Examiners must mark the first candidate in exactly the same way as they mark the last." / "Mark schemes should be applied positively. Candidates must be rewarded for what they have shown they can do rather than penalised for omissions." / "There is no ceiling on achievement. All marks on the mark scheme should be used appropriately." / "Crossed out work should be marked UNLESS the candidate has replaced it with an alternative response."

Two concrete, practical consequences a student should actually act on:

1. Don't erase — cross out, once, and move on. If you cross out a wrong attempt and don't replace it, the crossed-out work is still marked. Erasing entirely, or scribbling illegibly, denies the examiner something they are otherwise instructed to credit.
2. Multiple attempts at the same part are handled by an explicit rule, verbatim: "If all but one attempt is crossed out, mark the attempt which is NOT crossed out. If either all attempts are crossed out or none are crossed out, mark all the attempts and score the highest single attempt." The practical consequence: if you're unsure which of two workings is right and are running low on time, leave both visible rather than crossing either out — the examiner takes your best one. The one thing that actively costs you is crossing out your better attempt while leaving a worse one standing alone.

#2.4 The Statistics-specific marking note

Printed once per scheme, under "Special notes for marking Statistics exams (for AAs only)," verbatim:

"Any correct method should gain credit. If you cannot see how to apply the mark scheme but believe the method to be correct then please send to review." / "For method marks, we generally allow or condone a slip or transcription error if these are seen in an expression. We do not, however, condone or allow these errors in accuracy marks."

This is already taught in depth in sampling-tree-diagrams-venn-diagrams.ts and normal-distribution-standardisation-and-conditional-probability.ts, so only the headline practical rule is repeated here: the structure of your method — which quantities you multiplied, which conditional-probability ratio you set up — is worth writing out visibly even under time pressure, because that structure is where the more forgiving mark lives. A transcription slip inside a correctly-set-up method can still earn the M mark; the same slip surviving into a final accuracy line generally cannot.

#2.5 The misread penalty

Verbatim: "For misreading which does not alter the character of a question or materially simplify it, deduct two from any A or B marks gained, in that part of the question affected." Verified in active use, Oct 2021 ER Q3(b): "an alternative mark scheme, applying the usual misread penalty, was used for parts (b) and (c)" after a student misread a count/roll figure. Already taught in show-that-answer-discipline.ts — repeated here only as the headline fact: misreading a given number doesn't zero a question outright if your method from that point on is otherwise correct, but it does cost you a fixed 2-mark penalty on the marks affected, on top of whatever the misread number itself cost you.

---

#3. Statistics-specific conventions

#3.1 Correct use of statistical notation

Verified against the spec's own "Notation and formulae" box and the formula booklet (spec pp. 13–17, Issue 2, Jan 2021), and against live mark-scheme behaviour:

What you must produce from memory (spec §S1.2 states outright: "Formulae that students are expected to know are given below and will not appear in the booklet"):

• $P(A') = 1 - P(A)$
• The independence-specific forms only: $P(B \mid A) = P(B)$, $P(A \mid B) = P(A)$, $P(A \cap B) = P(A)P(B)$
• $E(aX+b) = aE(X) + b$, $\text{Var}(aX+b) = a^2\text{Var}(X)$
• $F(x_0) = P(X \le x_0) = \sum p(x)$ for $x \le x_0$
• $Z = \dfrac{X - \mu}{\sigma}$
• Mean $\bar{x} = \dfrac{\sum x}{n}$ or $\dfrac{\sum fx}{\sum f}$; standard deviation; $IQR = Q_3 - Q_1$

What is given in the yellow formula/statistical-tables booklet (so does not need to be memorised, but does need to be found and used correctly under time pressure): the general, non-independent form $P(A \cap B) = P(A)P(B \mid A)$; the full Bayes'-ratio form of $P(A \mid B)$; the base definitions $E(X) = \sum x_i P(X=x_i)$ and $\text{Var}(X) = \sum(x_i-\mu)^2 P(X=x_i) = \sum x_i^2 P(X=x_i) - \mu^2$ (not the $aX+b$ shortcut forms); the full $S_{xx}$, $S_{yy}$, $S_{xy}$ formulae (both deviation and computational forms), $r = S_{xy}/\sqrt{S_{xx}S_{yy}}$, $b = S_{xy}/S_{xx}$, $y=a+bx$ where $a=\bar y - b\bar x$; and both statistical tables (§3.3 below).

A genuinely useful rule of thumb, stated directly in WST01-verified-facts.md §2a: "if it's a general formula with a summation or a table, it's on the yellow sheet; if it's a rule for simplifying a special case, you need to know it." Independence, linear transformations, standardisation and IQR are all "special-case simplifications" in that sense — they are exactly the formulae this paper expects memorised.

Notation precision that costs real marks, not stylistic pedantry:

• Unsimplified fractions can be A0. Verified, Jan 2021 Q1(c) mark scheme: a correct conditional probability left as $0.15/0.4$ is explicitly not credited — it has to be reduced to $3/8$ or $0.375$ to earn the mark. oe covers equivalent forms of a correctly-simplified value; it does not cover an unsimplified fraction standing in for the same thing.
• Fractions are not accepted in a final regression equation. Verified, Jun 2022 Q2(c) ER, verbatim: "it is also important to note that fractions are not accepted in a final regression equation; the values of a and b are both estimates and so a fractional answer is not appropriate." Already taught in regression-gradient-interpretation-extrapolation.ts and correlation-coefficient-and-regression-calculation.ts — restated here because it is a genuine notation rule, not just a style preference: decimal form only, on the final line.
• Blank regions on a Venn diagram are never assumed to be zero. Verified, Jun 2022 Q4(c) ER: "some candidates left out the 0 in the outside region of N and should be reminded that blank spaces are not assumed to be 0s in Venn diagrams." Write the 0 in.
• $r$ is invariant under linear coding — a real, quotable, testable property of the notation itself, not just a calculation shortcut: "r not affected by (linear) coding" (Jan 2024 MS Q2(d)). If a question codes the data and later asks for $r$ on the original variables, recomputing from scratch is wasted time, not extra rigour.

#3.2 "Conclusion in context" — corrected for what this paper actually is

The task brief behind this document asked specifically about hypothesis-test structure — $H_0$/$H_1$, test statistic, critical value, and a conclusion given in context — with an instruction to check whether this paper's real examiner reports confirm that framing. They do not, because the framing does not apply: WST01 (Statistics 1) contains no hypothesis testing at all. This is confirmed twice, directly in the spec's own guidance notes transcribed in WST01-verified-facts.md §1: item 2.2 (measures of location) states outright "significance tests will not be expected", and item 4.3 (the product moment correlation coefficient) states "derivations and tests of significance will not be required." A full-text search of the facts bank for "hypothesis," "H0," "H1," "critical value," or "test statistic" returns nothing beyond that one guidance line. Hypothesis testing (the binomial test, and later the correlation-coefficient significance test) belongs to S2/S3 in the Edexcel IAL Maths suite, not S1. This is a genuine, load-bearing correction, not a missing feature of this document — writing $H_0$/$H_1$ convention rules for WST01 would be fabricating content for a topic this paper does not examine, which the task's own hard rules forbid.

What the real WST01 equivalent of "conclusion in context" actually is. The paper carries an explicit AO4 weighting — "comprehend translations of realistic contexts into mathematics; use results to make predictions/comment on context" (5–10 of 75 marks, spec pp. 68–69) — and this course's own research pass independently identifies "a worded interpretation/comment in context" as the single most consistently examiner-flagged weak point across the whole facts bank. Concretely, this paper's real "give your conclusion in context or lose the mark" moments are:

1. Interpreting a regression gradient. A bare direction statement is not credited. Verified, five separate series, five distinct concrete slips: mis-scaling the units ("a rise in the number of employees led to a rise in the amount spent on paper of $156 and not $1.56", Oct 2021); stopping at correlation language instead of a rate ("only the most able candidates successfully managed to write that the GDP increases by 31.2 billion dollars for every 1 million increase in the population", Jun 2022); mixed units ("some students mixed up the units (grams and °C)", Jan 2023); reversed direction (Jun 2024). The credited answer form is specific: name the rate, name both variables, get the units and scale right.
2. Judging the reliability of an extrapolated prediction. Verified, Jun 2022 Q2(e) and Jan 2023 Q6(c): a bare "unreliable" is worth nothing — "it was rare to see responses which accurately assessed the reliability of the estimate... many stated the estimate was unreliable, but they were unable to refer to the correct variable or value that is not in the range." The credited form names the specific value and the specific boundary it falls outside of. Where a question gives no min/max at all — a real, confirmed gap this course's own audit found and closed in regression-gradient-interpretation-extrapolation.ts — the fallback is comparing the predicted $x$ to $\bar x$ (or the spread implied by $S_{xx}$), not falling back to a generic "extrapolation so unreliable," which the real mark scheme explicitly zero-credits ("'Extrapolation so unreliable' on its own is B0", Jun 2022 Q2(e) marking note).
3. Comparing two distributions in words. Verified, three series, consistent pattern: "surprisingly too many students failed to give supporting figures... a question like this will require some context..., a reference to a named statistic and supporting figures" (Jun 2024); "few comments referring to the distribution of ages were seen... others commented on the similarities [when asked for differences]" (Oct 2021); and a specific, chronic misconception — "A very common misconception was that the mean is more accurate than the median... showing no appreciation that the mean is just one measure of average and the median is another" (Jun 2022). The credited answer form again is specific: name a statistic, give its value for both groups, and answer the actual question asked (difference vs similarity).
4. Assumptions and refinement in mathematical modelling (spec 1.1). The facts bank flags this sub-topic as thin in the past-paper record — no standalone question exists — but AO3 (15–20 of 75 marks, unusually high for this qualification) is explicitly "recall/select/use knowledge of standard mathematical models... including assumptions and refinement," meaning modelling-in-context marks arrive embedded inside other questions rather than as their own item.

The rule to internalise: on this paper, "conclusion in context" never means citing a critical value against a test statistic — it means naming a specific statistic, value, or rate, attaching the right units and scale, and connecting it explicitly to the real-world quantity the question is about. A technically true but generic sentence ("there is a positive correlation," "the estimate is unreliable," "the mean is the average") is worth nothing on any of these four question types, verified independently across five different examiner reports.

#3.3 Calculator-value vs formula/table-derived-value conventions

This maps directly onto AO5 — "use calculator technology/formulae booklet/tables accurately and efficiently; know when not to use them; give answers to appropriate accuracy" (5–10 of 75 marks) — and is grounded in the front page's own rounding instruction, verbatim, present on every question paper reviewed:

"Values from the statistical tables should be quoted in full. If a calculator is used instead of the tables, the value should be given to an equivalent degree of accuracy. Inexact answers should be given to three significant figures unless otherwise stated."

What this actually costs students, verified across three independent series, same underlying mechanism each time — a rounded table value substituted for the exact 4-d.p. one:

• Jan 2021 Q3(b): "many students were using a z value of 1.03 or 1.04 rather than the value 1.0364 from the 'Percentage Points of the Normal Distribution' table and they lost a mark for the reduced lack of accuracy... some students are opting to use calculators for this sort of question and invariably they fail to show sufficient accuracy."
• Oct 2021 Q6(b): "not using an inaccurate value such as 1.64 instead of 1.6449."
• Jun 2024 Q5(b): "students should be reminded that when values are required from the tables, they need to be 4 decimal places. A common error was to use z value = 0.25" rather than the precise equivalent.

A second, real convention this document adds because it is genuinely useful and does not yet appear consolidated anywhere in the curriculum: on an ordinary "find" part (as opposed to a "show that" part), a bare correct final answer can score full marks — verified, Jan 2021 Q3(a) mark scheme: "Ans only 3/3" for the value 0.0692–0.0694. But this allowance is itself precision-conditional, on the same question: part (b) of the same mark scheme states a bare "23.6" scores only M1B0A1 (losing the middle B mark), while a bare "awrt 23.60" scores the full 3/3. The lesson: skipping working is only ever safe if the value you write down already carries the precision the mark scheme demands — 23.6 and 23.60 look identical in ordinary usage but are worth different marks on this exam, because only the second one proves the calculation, not a rounding, produced that final digit.

A confirmed but explicitly non-universal alternative: on the Jan 2023 Q1(d) mark scheme, the standard deviation formula's $/n$ form (population SD, giving awrt 7.26) and a $/(n-1)$ "sample standard deviation" form (giving a different rounded answer, awrt 7.3) were both credited for the same 2 marks. This is a real, verified alternative on one specific series — checking the June 2023 mark scheme alongside it found no equivalent credit, so this is documented here as a confirmed one-series fact, not a rule to rely on generally. The safe default remains the $/n$ population form the spec expects reconstructed from memory.

---

#4. Self-mark checklist

Run this against your own script before checking any mark scheme. Every line traces to a verified mark-scheme convention or examiner-report pattern above; none is a generic study-skills platitude.

Notation and precision

• Every table value you copied down has 4 decimal places, not a rounded 2- or 3-d.p. version (§3.3) — check this specifically anywhere you used the Percentage Points table to find an unknown $z$.
• Every fraction in a final probability answer is simplified, not left as an unreduced ratio like $0.15/0.4$ (§3.1).
• Your final regression equation has no fraction in it — decimal form only (§3.1).
• Every region on a Venn diagram you drew has a value in it, including any region that's genuinely zero — a blank space is never read as 0 (§3.1).
• Any inexact final answer is given to 3 significant figures, unless the question states otherwise (§3.3, front-page rounding instruction).

"Show that" discipline (full treatment: show-that-answer-discipline.ts)

• Every "show that" line shows the intermediate values, not just the given target repeated back — a bare "= 0.15" with no standardisation or table step visible earns nothing, even if it's correct.
• Your working carries one extra decimal place or significant figure beyond the value printed in the question, so it proves the answer was derived rather than rounded to match.
• If the question says "show that N outliers exist" (or any show-that-a-count question), you have actually listed the N values, not just stated the boundary that would identify them.

"Conclusion in context" — the real WST01 version (§3.2)

• A regression-gradient interpretation names the rate, both variables, and the correct scale/units — not just "as $x$ increases, $y$ increases."
• A reliability judgement on an extrapolated prediction names the specific value and the specific range boundary (or, if no range is given, compares to $\bar x$) — never a bare "unreliable" or "extrapolation so unreliable" on its own.
• A comparison between two distributions names a specific statistic with its value for both groups, and answers the actual question asked (difference vs similarity, not the reverse).
• You have not claimed the mean is more accurate than the median (or vice versa) — they measure different things; neither is intrinsically "more accurate."

Mark-scheme mechanics (§2)

• Where you're unsure between two workings for the same part, you left both visible rather than crossing one out — the examiner marks your best attempt, but only if it's still there to see.
• Wrong working genuinely crossed out (not scribbled over) is fine — but check whether the specific question overrides isw (rare, but real — Jan 2024 Q2(c)) before assuming later working can't cost you a mark.

Topic-specific self-checks that recur across the facts bank

• For any discrete probability function: do your probabilities sum to 1? (Missing $x=0$ from a domain is the single most-quoted error in this exact category, Oct 2021 Q4(d).)
• For any $\text{Var}(aX+b)$ calculation: did you square the constant $a$? (Confirmed independently three times across a three-year span.)
• For any outlier-fence calculation: did you use the rule given in the question, not a remembered $1.5 \times IQR$ default — and does your lower fence subtract the margin from $Q_1$, not add it?
• For any box plot: are your whiskers drawn to the real data value, not the calculated fence? (The real mark scheme is inconsistent about crediting the fence — safest default is always the real value.)

---

#5. Common-error digest

This section is a cross-referenced index, not a duplicate of the trap-taxonomy items already written into the 13 curriculum lessons — every item below either points at where the full, worked treatment already lives, or is flagged explicitly as new material added by this document specifically because it does not yet appear anywhere in the curriculum.

#5.1 Already fully covered in the curriculum (pointer only)

#5.2 Genuinely new to this document (not present in any curriculum lesson as of this pass)

1. Positive-marking principles and the crossed-out-work rule — §2.3 above. Verified verbatim from the mark schemes' own "General Marking Guidance," but this exact administrative-technique point (don't erase; leave ambiguous multiple attempts visible rather than crossing out your best one) does not appear in any of the 13 existing lessons.
2. The multiple-attempts scoring rule — same source, same gap. "If all but one attempt is crossed out, mark the attempt which is NOT crossed out. If either all attempts are crossed out or none are crossed out, mark all the attempts and score the highest single attempt."
3. The per-question timing table for one real series (Jan 2023), reproduced in §1 above from paper-anatomy.ts's own research, but reframed here specifically as an exam-day pacing tool rather than a structural fact — a genuinely new synthesis (not new data) this document adds.
4. The consolidated "conclusion in context" framing of §3.2 — the four real question types where this paper actually demands a specific, contextualised statement (gradient interpretation, reliability judgement, distribution comparison, modelling assumptions) are each separately verified and separately taught in their own lessons, but no single place in the curriculum states the general pattern across all four as one exam-technique rule. This document is the first to do so.

---

#6. Exam-day protocol

Grounded entirely in verified front-page instructions and repeated examiner-report advice — nothing below is generic exam wisdom invented for this document.

1. Check the question count before you start. The front page states it directly ("There are N questions in this question paper") — it has been 6, 7 or 8 across the 14 series reviewed. Don't assume 6.
2. Bring an ordinary scientific calculator, not a CAS. Verified, spec Appendix 6: calculators must not have "the facility for symbolic algebra manipulation, differentiation and integration, or [have] retrievable mathematical formulae stored in them," and must not communicate with other machines. Calculators are allowed on every unit of this qualification, including S1 — this is not a non-calculator paper.
3. You will be given the yellow "Mathematical Formulae and Statistical Tables" booklet — named verbatim, by colour, on every front page. Know its contents cold (§3.1) so you're not searching for a formula under time pressure that you were actually expected to know from memory.
4. Budget against each question's own printed mark allocation, exactly as the front page instructs — "use this as a guide as to how much time to spend on each question" — not against a flat 72 seconds/mark applied blindly. Expect the Normal-distribution question specifically to run long (§1).
5. Draw a diagram on any Normal-distribution question, even when not asked to. This is Pearson's own repeated, explicit recommendation, not this document's advice: "A simple diagram would have helped many to avoid this error" (Jan 2023 Q5(a), on the subtract-from-1 confusion); "an error that could probably have been avoided if they had drawn a suitable diagram" (Jan 2021 Q3(b), on a sign error). A sketched normal curve with the relevant region shaded catches both of the two most commonly confirmed Normal-distribution errors before you commit to an answer.
6. On every "show that" item, write one extra decimal place or significant figure beyond the printed target, and show the intermediate step, not just the standardisation and the final line (§4).
7. Answer every question — compulsory, no choice, confirmed on both the spec and every front page. A partially-attempted late question with a diagram and a labelled method line still earns real M and B marks even if you run out of time before the final accuracy line.
8. Quote table values to 4 decimal places; give inexact final answers to 3 significant figures, unless the question states otherwise — the front page's own rounding instruction, and the single most-repeated accuracy trap in the whole facts bank (§3.3).
9. Don't erase ambiguous working — cross it out once if it's wrong, or leave two genuine attempts both visible if you're unsure which is better (§2.3, §4).

---

#7. Genuine gaps in this document — stated plainly, not silently omitted

• No hypothesis-testing content exists in this document, because none exists on this paper. See §3.2. If a future task brief for a different Statistics unit (S2 or S3) asks for the same H0/H1/critical-value/conclusion-in-context framing, that framing genuinely applies there — it does not apply to WST01, and this document does not pretend otherwise.
• The per-question topic breakdown in §1's timing table is incomplete for Q3 and Q4 of the one real series it draws on (Jan 2023) — paper-anatomy.ts, this document's source for the mark allocation (11 and 13 marks respectively), did not itemise those two questions' actual topics as part of its own research pass. This document infers "likely discrete random variables" for both from a separate, independent citation elsewhere in the facts bank (Jan 2023 Q3(a) and Q4(i)(a), cited as "show that" algebra examples under spec Section 5) — an inference, not a confirmed full topic breakdown, and flagged as such rather than stated as fact. Q6's topic (regression) is confirmed directly from two separate facts-bank citations (Q6(a), Q6(c)) and is not a gap.
• Examiner-report coverage across the whole facts bank remains 6 of 14 mark-scheme series (Jan 2021, Oct 2021, Jun 2022, Jan 2023, Jun 2024, and a 2026-09-25 spot-check of Jun 2019) — the facts bank itself states this explicitly as a lower bound, not exhaustive coverage, and that caveat is inherited here rather than smoothed over. Eight series' examiner reports (Jun 2021, Jan 2022, Oct 2022, Jun 2023, Oct 2023, Jan 2024, Oct 2024, Jan 2025) were searched for but not conclusively located in the original research pass. The two genuine gaps the one-series 2026-09-25 spot-check surfaced (an amended-data-value box plot; a positive-E(X)-vs-P(win) conflation) both came from a single previously-unread series, which is itself evidence more gaps likely remain in the eight still unread — this document's common-error digest (§5) should be read with the same caveat as the facts bank's own trap taxonomy: a lower bound, not a finished inventory.
• The Jan 2023 "n−1 sample standard deviation" alternative-credit finding (§3.3) is confirmed on one series only — checking June 2023's mark scheme alongside it found no equivalent credit. This document deliberately does not generalise it into a universal WST01 convention.
• This document performed no new PDF fetches or independent primary-source verification of its own — every claim above traces either to WST01-verified-facts.md's own already-verified record, or to the already-audited curriculum .ts files, both cited by series and question number throughout. Where this document adds synthesis not present in either source (§2.3's two practical rules; §3.2's four-question "conclusion in context" framing; the timing table's reframing as an exam-day tool), that synthesis is built from verbatim-quoted material already independently verified elsewhere in this course, not from new research.