Statistics 1

Mock Paper Protocol

#WST01 (Statistics 1) — Mock Paper Protocol

Provenance: written 2026-09-29, built entirely on research/veridian/WST01-verified-facts.md (compiled 2026-09-12, spot-check addendum 2026-09-25) and cross-checked against the four documents already live in this folder — forge-exam-technique-guide.md, forge-battle-card.md, codex-formula-and-method-bank.md, signal-frequency-and-boundaries.md — plus content/veridian/curriculum/wst01/paper-anatomy.ts and content/veridian/question-bank/wst01.ts. No new PDFs, web fetches, or independent primary-source verification were performed for this pass; every factual claim below traces to that already-verified record, cited by series and question number where the source document itself cites one. This document does not repeat the mark-code definitions, the notation rules, the per-topic error digest, or the exam-day protocol those four documents already give in full — it points at them instead. What's genuinely new here is the process: how to select a real paper, sit it under real timing discipline, and self-mark it honestly against this paper's own real mark-scheme conventions, then turn the result into a revision decision rather than a mood.

#What this document is, and isn't

This is not a content lesson (that's the 13 files in content/veridian/curriculum/wst01/), not a condensed technique reference (forge-exam-technique-guide.md), not a formula bank (codex-formula-and-method-bank.md), and not a last-look morning-of page (forge-battle-card.md). It is the missing fifth piece: a protocol for the specific, repeatable act of sitting one of the 14 real past papers under mock conditions and then marking it against the real scheme without quietly being kind to yourself. Where a step below needs a mark-code meaning or an error's full citation, it points at the document that already has it rather than restating it.

This course holds no local archive of the 14 real WST01 papers themselves — the facts bank's own §3 records that every source it used was fetched fresh from qualifications.pearson.com or its physicsandmathstutor.com / dynamicpapers.com mirrors, not stored here. Running the protocol below means sourcing a real question paper, mark scheme, and (where one exists) examiner report the same way the facts bank did, from the same kind of source — this document doesn't manufacture a paper to sit, and doesn't pretend one is bundled with it.

#1. Choosing which of the 14 real series to sit

Fourteen series exist in the verified record: Jun 2019, Jan 2021, Jun 2021, Oct 2021, Jan 2022, Jun 2022, Oct 2022, Jan 2023, Jun 2023, Oct 2023, Jan 2024, Jun 2024, Oct 2024, Jan 2025 (§2/§3 of the facts bank). Two real, checkable facts should decide the order you sit them in, not just "earliest first":

Only 6 of the 14 have a located, fully-read examiner report: Jun 2019, Jan 2021, Oct 2021, Jun 2022, Jan 2023, Jun 2024 (facts bank §3; signal-frequency-and-boundaries.md uses the same 6-of-14 denominator). The other 8 — Jun 2021, Jan 2022, Oct 2022, Jun 2023, Oct 2023, Jan 2024, Oct 2024, Jan 2025 — have a verified mark scheme but no examiner-report commentary behind it in this course's record. A mock is only as good as what you can learn from marking it, and an examiner report gives you real qualitative commentary (what a whole cohort actually did wrong on that exact paper) that a mark scheme alone cannot. Practical rule this document adds: don't burn all 6 ER-backed series early out of habit. Spread them across your mock cycle — sit at least one of the 6 near the start (to calibrate what real per-question feedback looks like) and keep at least one in reserve for later, when you're closer to your real sitting and the extra layer of commentary matters more.
The question count is 6, 7 or 8, and later series trend toward more, shorter questions (6: Jun 2019, Jan 2021, Jun 2021, Oct 2021, Jun 2022, Oct 2023, Jun 2024; 7: Jan 2022, Oct 2022, Jun 2023, Jan 2025; 8: Jan 2024, Oct 2024 — facts bank §2). If your mock cycle only ever uses early, 6-question series, you will never rehearse the front page's own "There are N questions" line actually surprising you. Deliberately include at least one 7-question and one 8-question series in your mock cycle, not only 6-question ones — the point of a mock is to rehearse the real range of what you'll see, not the most familiar member of it.
Do not sit the same series twice inside one course cycle. If you must re-sit one (e.g. to time an improved attempt after fixing a specific weakness), treat the second sitting as a technique drill, not a diagnostic mock — you already know the questions, so it cannot tell you anything honest about recall or timing under genuine uncertainty.

#2. Before you sit it: sourcing and setup

A self-run mock has no invigilator handing you materials, so you have to assemble the real exam conditions yourself, deliberately:

Source the actual question paper, mark scheme, and (if available) examiner report the same way this course's research pass did — official PDFs from qualifications.pearson.com, or the physicsandmathstutor.com / dynamicpapers.com mirrors that reproduce them unmodified (verified by the facts bank as carrying Pearson's own copyright footer and Publications Code, §3). Do not sit a mock from a source you can't verify is the real, unmodified paper — a garbled or edited copy would make every downstream timing and marking judgement in this protocol meaningless.
Get your own copy of the "Mathematical Formulae and Statistical Tables (Yellow)" booklet before you start, not partway through. On the real exam it's handed to you; in a self-run mock, if you go looking for it mid-paper you've just measured your search time, not your statistics.
Check your calculator against the real restriction, not from memory: no symbolic algebra, no symbolic differentiation/integration, no stored formulae or text, no communication with other machines (spec Appendix 6, facts bank §2). An ordinary scientific calculator that happens to violate one of these is not neutral — using it would make your mock easier than the real thing in a way you won't be able to repeat on the day.
Set a single 90-minute timer for the whole paper, not a per-question one. The front page's own instruction is to use each question's printed mark total as your own guide to pacing — "use this as a guide as to how much time to spend on each question" (facts bank §2) — not to have an external per-question alarm doing that judgement for you. The skill being tested by a mock includes deciding, yourself, in real time, whether to keep going on a question or move on.
Have nothing else open. No lesson .ts files, no this-document's own checklists, no worked examples. The entire diagnostic value of a mock is that it tells you what you can actually produce unaided; consulting notes mid-mock converts the exercise into an open-book rehearsal and silently invalidates every timing and error-log conclusion drawn from it afterward.

#3. Sitting it: measure your real time, don't estimate it afterward

forge-exam-technique-guide.md §1 already gives the real timing budget — 72 seconds/mark on average, and a real per-question breakdown for one series (Jan 2023: 10, 10, 11, 13, 17, 14 = 75 marks) showing the Normal-distribution question running close to a quarter of the whole paper on its own. That is a plan. What a mock adds is a measurement against that plan, and a measurement is only honest if you record it as you go, not reconstruct it from memory once you've finished:

Write the clock time (or elapsed minutes) in the margin the moment you finish each question, before moving to the next one. Five numbers scribbled in real time beat a confident but reconstructed guess made after the fact about "roughly how long Q3 took."
Do not stop and go back to fix an earlier answer mid-paper unless you would genuinely be allowed to on the real exam (you are — nothing stops you turning back a page) — but if you do, note that you went back, because it will otherwise distort your per-question timing record and hide exactly the behaviour (second-guessing, re-checking) worth knowing about.
If you run out of the full 90 minutes with questions unattempted, stop anyway and mark what you'd have written with a different-coloured pen for another 10–15 minutes. This gives you two real numbers instead of one blurred one: your genuine 90-minute score, and how much more you'd have scored with the time you didn't have — which is a different, and separately useful, diagnostic (see §7).

#4. Immediately after finishing, before you look at the mark scheme

This step has to happen in this order, or it cannot happen honestly at all: once you've seen the mark scheme, you can no longer tell the difference between "I knew this and made a slip" and "I am now reading the correct method into my own vaguer working after the fact." So, pen still in hand, mark scheme still closed:

Go back through your own script and put a small mark next to any answer you know you guessed, fudged, or wrote down without being sure of the method — not because it's necessarily wrong, but because you want an honest record of where you felt uncertain, independent of whether the mark scheme later agrees with you.
Circle every "show that" part where you wrote the target value with less working than you'd give an ordinary "find" part. This paper's mark schemes distinguish sharply between the two — a bare correct final answer can legitimately score full marks on an ordinary "find" (verified, Jan 2021 Q3(a) mark scheme: "Ans only 3/3"), but the same bareness on a "show that" line earns nothing, however correct the number is, because there is no method mark to chain an accuracy mark to (the mechanism show-that-answer-discipline.ts and forge-exam-technique-guide.md §4 both already teach in full). Flagging this now, before marking, tells you honestly whether you actually wrote the intermediate step or are about to convince yourself you did once you see what was required.
Note which questions you didn't reach at all, separately from questions you attempted and got wrong. These are different failure modes (recall/skill gap vs pacing) and get merged into one number if you don't separate them before marking.

Only once this pass is done, open the mark scheme.

#5. Self-marking: applying the real conventions to your own paper honestly

forge-exam-technique-guide.md §2 and forge-battle-card.md §2 already give the full, verbatim meaning of every mark code this paper's schemes use (M, A, B, ft, cao, cso, awrt, isw, oe, SC, bod, dep/indep) — this section does not repeat those definitions. What it adds is the procedure for applying them to your own work without quietly marking generously, which is the single biggest way a self-run mock produces a false, comfortable score.

#5.1 Mark in two passes, not one

First pass: go through mechanically, matching your final answers against the mark scheme's stated values or tolerance (cao, awrt) only. This gives you a rough raw score fast, and nothing more — don't trust it as your real result yet.

Second pass: go back through every part and check the actual M/A/B chain the mark scheme specifies, not just whether your final number matches. This is where self-marking leniency actually happens, and it is worth naming the specific, real ways it happens on this paper rather than trusting yourself to notice generically:

"M0 A1 is impossible" applies to you too. If your working doesn't show the method the scheme credits — verbatim, all 14 schemes reviewed, "M0 A1 is impossible" (facts bank §4) — you cannot award yourself the accuracy mark just because your final number happens to be right. A right answer with an invisible or wrong method is a right answer with zero marks on this paper's real convention, and a self-marker who gives themselves the accuracy mark anyway is not measuring what the real exam would give them.
An unsimplified fraction is not "basically right." Verified, Jan 2021 Q1(c): a correct conditional probability left as $0.15/0.4$ scored A0 — it had to be reduced to $3/8$ or $0.375$ . If your own script has an unreduced fraction sitting where a final answer should be, mark it as the real scheme would: wrong, not "nearly right."
A rounded table value is a real, quantified way self-markers let themselves off. This paper's own examiner reports confirm the exact same slip three separate times across a three-year span — 1.03/1.04 used instead of the table's 1.0364 (Jan 2021 Q3(b)); 1.64 instead of 1.6449 (Oct 2021 Q6(b)); 0.25 instead of the precise equivalent (Jun 2024 Q5(b)). If your working shows a rounded z-value where the scheme wants 4 decimal places, that line loses the mark on the real paper — mark it that way on your mock, even though the final answer often "looks" fine.
Watch for reasoning that matches the question's own wrong logic, not the mark scheme's. The clearest real example of this trap: June 2019 Q5(e) gives a fictional character who says "because E(X) > 0 he will play the game" and asks whether you agree. The mark scheme requires citing $P(\text{win}) < 0.5$ as the deciding quantity, not $E(X)$ — verbatim, "Amar should not play the game" is only credited via the $P(\text{win})$ route. The examiner report confirms this was the dominant failure, not a rare one: "many attempts focussing on the fact that E(X) > 0" (June 2019). If you find yourself, while self-marking, tempted to accept your own answer because it "sounds reasonable," check it against the actual quantity the scheme names — not against whether your reasoning sounds plausible in isolation. (The full worked question, with the complete mark scheme and examiner-report text, is in content/veridian/question-bank/wst01.ts.)
Check isw isn't quietly protecting a mistake it doesn't cover. The default is forgiving, but it is a default, not a guarantee — verified real override, Jan 2024 MS Q2(c): "do not isw if any further calculation is done after multiplying by 9/5." If your own further working after a correct line changed the answer, check whether that specific mark point is one of the ones that switches isw off before assuming it protects you.
Apply the misread penalty to yourself if it's genuinely earned — deduct 2 from the A/B marks in the part affected, the same real penalty the scheme applies to a candidate who misreads a given number without materially simplifying the question (facts bank §4; live example, Oct 2021 Q3(b)). Don't zero the whole part, and don't quietly ignore the deduction either.

#5.2 When the mark scheme is genuinely ambiguous, mark against yourself

The real Statistics-specific marking note tells an actual examiner what to do when they can't see how a mark point applies: "please send to review" (facts bank §4). You have no review process. The honest substitute, stated plainly because the facts bank doesn't state it for you: on any mark point where you can't tell whether your working satisfies the scheme, mark it as not earned. A mock's entire value is as a diagnostic of real gaps; inflating an ambiguous call in your own favour doesn't just flatter one paper's score, it hides the exact thing you're running the mock to find.

#5.3 Record the raw score, but don't stop there

Write down the raw mark out of 75 (or out of whatever the series totals — always 75 on this paper, per §2 of the facts bank). Then go straight to §6 below — the raw number on its own is the least useful output of a mock; the categorised error log is what actually changes your revision plan.

#6. Build an error log, categorised by why the mark was lost

Every dropped mark goes into exactly one of the following categories. These categories are not invented from nothing — each one corresponds to a real, recurring failure type this course's own research has independently verified appearing across multiple series (full citations: forge-exam-technique-guide.md §5, forge-battle-card.md §4):

Conceptual gap — you didn't know the method at all, not even an attempt. This is a revision gap, not a technique gap; it means going back to the relevant curriculum lesson, not this document.
Notation/precision slip — you knew the right method but lost the mark on execution: a rounded z-value instead of 4 d.p., an unsimplified fraction, a missing $\sqrt{}$ in a PMCC denominator (Jun 2024 Q4(b)), forgetting to square the constant in $\text{Var}(aX+b)$ (confirmed independently three times — Jan 2021 Q4, Jun 2024 Q2(c), Q3(d)), a fraction left in a final regression line (Jun 2022 Q2(c)), a blank Venn region not written as 0 (Jun 2022 Q4(c)).
"Show that"/context-answer discipline — you had the right method but didn't write enough of it down to earn the chained accuracy mark, or gave a bare qualitative "conclusion in context" answer where the scheme wants a named rate, value, or statistic (the four real trigger points are fully catalogued in forge-exam-technique-guide.md §3.2 and forge-battle-card.md §3 — don't re-derive them here, just tag the error against that list).
Time pressure — you either didn't reach the question (confirmed directly from your §3 margin timestamps and §4's separated "didn't reach" list) or rushed a late question visibly compared to an earlier one. This category only exists if you actually recorded real elapsed times in §3 — without that, "time pressure" becomes a guess, not a diagnosis.
Administrative/carelessness — a genuine arithmetic slip, a misread number, or a transcription error inside an otherwise-correct method — the kind of error the Statistics-specific marking note explicitly says examiners will condone in a method mark but not in an accuracy mark (facts bank §4). If it recurs across more than one mock, it stops being "carelessness" and becomes category 2.

For each dropped mark, write: the question/part, the category number, and one line naming the specific thing that went wrong (not "got it wrong" — "used 1.64 instead of the table's 1.6449," or "wrote 0.15/0.4 unsimplified"). A raw score with no categorised log tells you that you lost 12 marks; a categorised log tells you which of the five things to actually go and fix.

#7. Reading the log against what this paper actually weights

Once you have two or three mocks' worth of categorised logs, look for where your category-1 (conceptual) errors cluster, and check that cluster against the real topic-frequency table in signal-frequency-and-boundaries.md §1:

Section 2 (data representation/summary) is confirmed tested in 6 of 6 fully-read examiner-report series — the only topic with that record. A conceptual gap here is a near-certain recurring cost, not a one-off.
The Normal distribution (6.1) is confirmed in 5 of 6, and is the single heaviest question wherever a full mark breakdown exists (17 of 75, Jan 2023) — a conceptual gap clustering here costs disproportionately more than the same number of category-1 errors spread across a lighter topic.
Discrete random variables (5.1–5.4) are also confirmed in 5 of 6.
Probability (3.1–3.4) and correlation/regression (4.1–4.3) are each confirmed in 4 of 6.
Mathematical modelling (1.1) has no standalone question anywhere in the 14-series record — a category-1 gap "here" almost never shows up as its own missed question; it shows up embedded inside AO3 marks on other questions (see forge-exam-technique-guide.md §3.2, point 4).

This is the one place in this protocol where a mock's pattern across repeats matters more than any single sitting's score: one mock's category-1 cluster could be noise; the same cluster appearing across two or three mocks, on a topic this table already shows is heavily and consistently examined, is a real, actionable revision priority.

#8. What a mock score genuinely cannot tell you

signal-frequency-and-boundaries.md §2 records, directly: a search of the facts bank for "grade boundary," "boundaries," and "raw mark" returns zero matches — no real Pearson grade-boundary data for WST01 exists anywhere in this course's verified record. This document does not manufacture one either. Two concrete consequences for what you do with a mock score:

Do not convert a raw mock mark into a predicted grade or a percentage-likelihood statement. There is no verified boundary to convert it against, and inventing one — even a "rough" one — is exactly the fabrication this course's standing discipline forbids. If a grade estimate matters to you, it has to come from Pearson's own published grade-boundary document for a specific series, fetched and read directly, not from anything in this protocol.
The one number a mock series genuinely licenses is a trend in your own raw marks and your own categorised-error counts over time, sitting after sitting — did category 2 (notation/precision) shrink after you started deliberately writing 4 d.p. table values? Did category 4 (time pressure) shrink after two mocks of margin-timestamping? That comparison is real and useful. A single mock's raw score treated as a grade prediction is not.

#9. A mock cycle, not a single sitting

One mock tells you about one paper. The real value comes from a short sequence, deliberately sequenced using the facts already established above:

Early in the cycle: sit a 6-question series with a located examiner report (e.g. Jan 2021 or Jun 2022) — the report gives you real qualitative commentary to check your own error log against, which calibrates whether your categorisation in §6 is actually matching what real examiners saw.
Mid-cycle: sit at least one 7-question and one 8-question series (Jan 2022/Oct 2022/Jun 2023/Jan 2025 for 7; Jan 2024/Oct 2024 for 8) — this rehearses the real structural variance this paper carries, not just its content.
Late in the cycle, closer to your real sitting: sit one of the remaining ER-backed series held in reserve (§1) — by this point your error log has real history behind it, so a full qualitative examiner report on this specific attempt is the most useful feedback you'll get from any single mock in the cycle.
Between sittings: don't just re-read your error log — go and actually work the specific category-1 and category-3 gaps it surfaced, in the curriculum lesson or technique-hub section that covers them, before the next mock. A mock cycle that never changes behaviour between sittings is just repeated measurement of the same gaps.

#10. Mock-day materials checklist

Distinct from forge-exam-technique-guide.md §6 (the real exam-day protocol) — this is what a self-run mock specifically needs that a real invigilated sitting hands you automatically:

The real, verified question paper and mark scheme for the chosen series, sourced from qualifications.pearson.com or a verified mirror (§2 above) — not a summary, not a rewrite.
Your own copy of the yellow "Mathematical Formulae and Statistical Tables" booklet, printed or open, before the timer starts.
A calculator checked against spec Appendix 6's real restrictions (§2 above).
A single 90-minute timer, plus a way to note elapsed time per question as you finish it (§3).
A second-coloured pen, in case you choose to keep going past 90 minutes to separately measure "what more I'd have scored with the time I didn't have" (§3).
Nothing else open — no lesson files, no this document, no earlier error logs (§2 above).
The mark scheme kept genuinely closed until the §4 cold self-flag pass is complete.

#Sources and limitations

research/veridian/WST01-verified-facts.md (primary, provenance-checked facts bank — §2 assessment structure, §3 source archive and the 6-of-14 examiner-report denominator, §4 mark-scheme conventions, §5 per-topic verified content, including the Jan 2021 Q1(c), Jun 2022 Q2(c)/Q4(c), Jan 2024 Q2(c)/(d), Jan 2021/Oct 2021/Jun 2024 rounded-z, and Jan 2021 Q3(a)/(b) "Ans only"/precision-conditional citations used above), and its June 2019 Q5 discrete-RV gap-closure entry (E(X)-vs-P(win), including the mark scheme and examiner-report text quoted in §5.1 above). content/veridian/technique-hub/wst01/ forge-exam-technique-guide.md, forge-battle-card.md, codex-formula-and-method-bank.md, and signal-frequency-and-boundaries.md (this paper's already-built layers — mark-code definitions, the four "conclusion in context" trigger points, the topic-frequency table, and the explicit finding that no grade-boundary data exists, all cross-checked here for no duplication rather than restated in full). content/veridian/question-bank/wst01.ts (the two full real worked questions — June 2019 Q2(c) and Q5 — cited above by series and part). content/veridian/curriculum/wst01/paper-anatomy.ts (the real per-question Jan 2023 mark breakdown and front-page pacing instruction, cited in §3 and §7).

This document performed no new PDF fetches, web searches, or independent primary-source verification — every specific factual claim above traces to one of the sources named in this section, cited by series and question number. Where this document adds something not present in any of them — the mock-selection sequencing in §1 and §9, the real-time timing-measurement protocol in §3, the before-the-mark-scheme cold self-flag pass in §4, the two-pass self-marking procedure and the "mark against yourself on genuine ambiguity" rule in §5.2, the five-category error-log taxonomy in §6, and the materials checklist in §10 — that is original procedural synthesis built on top of already-verified facts, not a new factual claim about the exam itself, and is presented as such rather than attributed to Pearson or to any source document. This document inherits the same two open caveats those sources already state rather than resolving them: examiner-report coverage remains 6 of 14 series (a lower bound, not exhaustive coverage — see WST01-verified-facts.md §3 and signal-frequency-and-boundaries.md), and no verified grade-boundary data exists for this paper at all (§8 above).