Statistics 1

Examiner Intelligence

#WST01 (Statistics 1) — Examiner Intelligence

Provenance: written 2026-09-29, built entirely on research/veridian/WST01-verified-facts.md (compiled 2026-09-12, spot-check addendum 2026-09-25). No new PDFs, web fetches, or external searches were performed for this document — the facts bank's own §3 is the only source used here, and its own honest denominator is used throughout: 6 of the 14 mark-scheme series have a confirmed, independently-read examiner report (Jun 2019, Jan 2021, Oct 2021, Jun 2022, Jan 2023, Jun 2024). Every "N of 6" count below uses that denominator, never 14. Cross-checked against forge-error-pattern-tracker.md, signal-frequency-and-boundaries.md, forge-exam-technique-guide.md and codex-formula-and-method-bank.md in this folder, and content/veridian/question-bank/wst01.ts, specifically to avoid restating what those already cover.

#What this document is, and isn't

The error-pattern tracker in this folder catalogues what goes wrong (which formula, which sign, which step). This document instead asks: how does Pearson's own report talk? Not the maths content of the mistake, but the recurring shape of the sentence reporting it — the phrases the examiner reaches for, where in the report a comment gets placed, and what that placement or phrasing implies about how a marker is actually reading a script. That is a genuinely different question from "what mistake," and it is one the other technique-hub documents don't ask.

The discipline this document holds to, stated plainly: a phrase or placement habit is only called a pattern below if it is confirmed in at least two of the six read series. Anything found in only one series is labelled a single finding and left as one, not stretched into a trend. No percentage or likelihood figure appears anywhere below — the source reports themselves never give one (see §3), so inventing one here would be worse than the reports being vague. Every count is "N of 6 series," nothing finer.

#1. The direct-address "should be encouraged / reminded / advised" register

Across four of the six series, the report shifts out of pure diagnosis ("candidates did X") into a distinct, imperative, forward-looking register that names a fix and addresses it straight at future students rather than at the cohort just marked:

Oct 2021 Q6(b): "Students should use the Percentage points table to find the z value when appropriate." The same series separately reminds candidates, on accuracy in general, that answers need "a minimum of 3 significant figures" unless the question states otherwise.
Jun 2022 Q4(c): "...should be reminded that blank spaces are not assumed to be 0s in Venn diagrams."
Jan 2023 Q1(d): "students should be encouraged to work with exact answers in their working of calculations."
Jun 2024 Q4(c): "students should be encouraged... to give answers to at least one more decimal place than the given value"; Q5(b) reminds candidates that table-derived values "need to be 4 decimal places"; and, in that series' general comments: "Students should be advised to read questions carefully as marks were dropped carelessly."

The construction is consistent enough to be worth naming: subject "students/candidates," modal "should," then either a passive past participle (be encouraged, be reminded, be advised) or a direct instruction (should use), always followed by a specific, executable action — not "revise harder" but "use the table," "keep exact values," "write four decimals." Read as a signal about how markers read scripts: the fix on offer is consistently procedural (a habit at the point of writing), never "know more content." That matches the mark-scheme's own M/A/B structure, where method is credited independently of the final figure — the examiner's advice and the mark scheme's design are pointing at the same thing, which is worth teaching explicitly rather than leaving implicit.

#2. The single-issue general comment, and what usually earns that slot

Four of the six series' reports use an opening general-comment paragraph — separate from the question-by-question notes — to name one specific weakness as that series' headline issue, before any question numbers appear:

Oct 2021: the general comment states, as a point applying beyond any single question, that many students do not really understand that "evidence has to be given" for 'show that' questions.
Jun 2022: "Candidates often assume independence when an appropriate conditional probability should be used instead."
Jan 2023: the report says students need to understand that "a full method should be given" for 'show that' questions.
Jun 2024: the general comment warns that marks were dropped carelessly for not meeting a question's demands, and singles out 'show that' questions specifically as needing "all the steps in the working to be shown." The same series separately opens its Normal-distribution commentary with "If asked to use standardisation then the standardisation should be shown."

Within those four, three (Oct 2021, Jan 2023, Jun 2024) spend the headline slot on the same specific thing: "show that" working discipline. That is not "examiners care about working" in the abstract — it is that when a WST01 report chooses to lead with one issue above the whole per-question breakdown, that issue is, on the evidence actually read, disproportionately likely to be this one. The other general-comment slot found (Jun 2022) went to conditioning-vs-independence instead — so the pattern is "a headline slot exists and gets used," with "show-that discipline" as its most frequent occupant among the four instances, not its only possible one.

#3. Vague quantifiers, and the absence of any real number

Every one of the six reports describes how many students made an error using words, never a count or a percentage: "the majority" (Jan 2023 Q1(a)), "a significant number" (Jan 2023 Q5(a)), "a small number" (Jun 2022 Q1(b)), "very few candidates" (Jun 2022 Q1(e)), "few candidates" (Jun 2019 Q2), "a surprising number" (Jan 2021 Q2), and, most often, unqualified "some" and "many." Two series go further and name the discriminating tier directly rather than the error rate: "only the most able candidates successfully managed to write..." (Jun 2022 Q2(d)) and "only the more able students scored full marks" (Jun 2024 Q5(c)) — both attached to the single hardest sub-step of a multi-part question (context-scaled gradient interpretation; a multi-stage inverse-Normal problem), which is a second, narrower two-series pattern in its own right: when a report uses "only the [most/more] able," it is flagging a genuine grade-discriminator step, not a routine slip.

What this means for using this document, and any document like it: there is no verified basis anywhere in the six read reports for converting "many," "a significant number," or "a small number" into an actual proportion. A revision resource that turns "many students" into "60% of students" or similar has manufactured a number Pearson never published. This document does not do that, and flags it here because the temptation is real and the source material makes it easy to notice: the reports themselves are consistently, deliberately unquantified.

#4. The same failure, named again years apart

Three mechanisms are each confirmed in more than one series, spaced far enough apart that the report is evidently re-diagnosing the same habit in a different cohort, not describing a one-off:

Rounded z-value instead of the table's 4-d.p. figure — Jan 2021 Q3(b) ("using a z value of 1.03 or 1.04 rather than... 1.0364"), Oct 2021 Q6(b) ("an inaccurate value such as 1.64 instead of 1.6449"), Jun 2024 Q5(b) ("the use of an inaccurate z value... a common error was to use z value = 0.25") — three series across a three-year span.
Subtracting from 1 in the wrong direction (either forgetting to, or doing it when not needed) — Jan 2021 Q3(a), Jan 2023 Q5(a), Jun 2024 Q5(a) — three series, and notably the error runs both ways across them (under-subtracting in some series' scripts, over-subtracting in others), which the report itself seems aware of: it recommends the same fix, "draw a diagram," each time rather than a direction-specific one.
Var(aX+b) computed without squaring the constant — Jan 2021 Q4(b–d) and, three years later, Jun 2024 Q2(c) and Q3(d) (two different questions in that one series). Two series, not three — forge-error-pattern-tracker.md in this folder already corrects an earlier three-series miscount of this exact mechanism; this document uses that corrected figure.

The implication worth stating plainly: within this six-series window, naming a mechanism in one report is not evidence it improves by the next series read. Nothing in the six reports suggests otherwise — each recurrence is reported in the same tone and register as if for the first time, not as "as noted previously." Treat every report's phrasing as a standing description of the paper, not a dated one.

#5. The recommended fix is almost always "make the working visible," not "learn more"

Pulling §§1, 2 and 4 together: across every distinct fix the six reports actually recommend — draw a diagram (Jan 2021 Q3(b), Jan 2023 Q5(a)), use the table's exact value (Oct 2021 Q6(b), Jun 2024 Q5(b)), show the standardisation step (Jun 2024, general), show full "show that" working (Oct 2021, Jan 2023, Jun 2024, general), check that probabilities sum to 1 before submitting (Oct 2021 Q4(d)) — none of them is "study this topic again." Every one is a habit executed at the moment of writing an answer down. That is consistent with what §1 already showed (the fixes offered are procedural, not content-based), and it is worth stating as its own point because it changes what "revising from examiner reports" should mean: not re-teaching the maths, but rehearsing the handful of habits the reports keep asking for.

#6. Single findings — real, but not generalised

Two things are worth recording precisely because they are not being claimed as patterns, in line with this document's own rule in the introduction:

Oct 2021 Q1(b), on independence vs mutually exclusive: "there was the usual confusion... highlighted by statements such as 'they are not independent as they overlap.'" The word "usual" is the report's own claim that this is chronic — but that claim appears in only one of the six reports read for this document, so it is recorded here as what it is: a single citation, whose chronicity is Pearson's assertion, not something this document has independently confirmed across series.
Jan 2021 Q3(b) names a specific cause the other five reports never repeat: it attributes the rounded-z-value error to students opting to use calculators for this type of question, who "invariably... fail to show sufficient accuracy." The rounded-z-value mechanism itself recurs three times (§4); the calculator attribution for it does not — it is named only once.

#Sources

research/veridian/WST01-verified-facts.md, specifically §3 (source archive and the 6-of-14 examiner-report denominator), §4 (mark-scheme conventions), §5 (per-topic examiner-report quotes, Sections 2–6), §6 (the "show that" cross-cutting pattern), §8 (flagged gaps), and §9 (the 2026-09-25 Jun 2019 spot-check). All direct quotations above are transcribed there from Pearson's own Principal Examiner Feedback PDFs (qualifications.pearson.com, cross-checked against Physics & Maths Tutor mirrors), and are reproduced here only as short quotes under Pearson's own copyright, not in bulk.