Statistics 1

Battle Card

#WST01 (Statistics 1) — Battle Card

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 three documents already live in this folder — codex-formula-and-method-bank.md (full formula notes and method templates), forge-exam-technique-guide.md (mark-coding system, notation rules, self-mark checklist), and signal-frequency-and-boundaries.md (topic-frequency table) — plus the 13 curriculum lessons at content/veridian/curriculum/wst01/, so nothing below repeats what those documents already explain in full. No new PDFs were fetched for this pass; every fact traces to the facts bank's own already-verified record. This is deliberately the most compressed of the four WST01 technique-hub documents — a last-look page for the morning of the exam, not a place to learn anything for the first time. Short quotes below are under ~15 words with series/question attribution, matching this course's standing no-fabrication, no-large-verbatim-reproduction rule; "seen in N of M series checked" phrasing is used throughout instead of any invented likelihood or grade-boundary figure, because none exists in the source record (confirmed in signal-frequency-and-boundaries.md §2 — no grade-boundary data was ever fetched for this paper).

#0. Exam vitals (tariff reminders)

75 marks, 90 minutes. Fixed on all 14 series reviewed (Jun 2019 → Jan 2025).
Question count is NOT fixed — 6 questions (7 series), 7 questions (4 series), 8 questions (2 series). Check the front page's own "There are N questions" line first.
All questions compulsory. No Section A/B/C, no MCQ anywhere in the real record.
Calculators allowed — ordinary scientific only (no CAS, no stored formulae, no machine-to-machine link). Not a non-calculator paper.
Yellow "Mathematical Formulae and Statistical Tables" booklet provided — named verbatim by colour on every front page.
Rounding rule, verbatim from every front page: "Inexact answers should be given to three significant figures unless otherwise stated." Table values: quote in full (4 d.p.), never rounded.
AO3 (modelling/interpreting in context) = 15–20 of 75 marks — unusually high for this qualification; this is why "conclusion in context" answers (§3 below) carry real weight, not a soft-skill afterthought.
Budget ≈72 sec/mark, but read each question's own printed mark total first — on the one series with a full verified per-question breakdown (Jan 2023), the Normal-distribution question alone was 17/75 marks, close to a quarter of the paper; the smallest was 10/75. Don't ration time evenly.

#1. Formula & definition cues — ultra-condensed

MEMORISE = spec-listed, will NOT appear in the booklet. GIVEN = printed in the yellow booklet. RULE GIVEN IN Q = no default to memorise; the question states it every time. Full notes, derivations and every cited trap: codex-formula-and-method-bank.md Part 1.

Topic	Cue	Status
Mean	$\bar x=\Sigma x/n$ or $\Sigma fx/\Sigma f$ (grouped: use midpoints)	MEMORISE
SD/variance	$\sigma^2=\Sigma fx^2/n-\bar x^2$ ; $\sigma=\sqrt{\sigma^2}$ — root wraps the whole thing, last	MEMORISE
IQR	$Q_3-Q_1$	MEMORISE
Quartile interpolation	locate class, interpolate proportionally; $n$ - or $(n{+}1)$ -based both tolerated	technique, no fixed formula
Outlier fences	lower $=Q_1-k\cdot IQR$ ; upper $=Q_3+k\cdot IQR$	RULE GIVEN IN Q
Coding decode	mean reverses the coding step; variance scales by the square, opposite direction	technique
$P(A')=1-P(A)$		MEMORISE
$P(A\cup B)=P(A)+P(B)-P(A\cap B)$		GIVEN
$P(A\cap B)=P(A)P(B\mid A)$ (general)		GIVEN
$P(B\mid A){=}P(B)$ , $P(A\cap B){=}P(A)P(B)$ (independence-specific)		MEMORISE
Tree/Venn	multiply along a path, add across paths; branches from a node sum to 1; without replacement: both count and total shrink each draw	technique
$S_{xx},S_{yy},S_{xy}$ (deviation + computational forms)		GIVEN
$b=S_{xy}/S_{xx}$ ; $a=\bar y-b\bar x$ ; $\hat y=a+bx$	final line: decimal, no fraction	GIVEN
$r=S_{xy}/\sqrt{S_{xx}S_{yy}}$	$-1\le r\le1$ ; invariant under linear coding	GIVEN
$p(x)=P(X=x)$ , $\Sigma p(x)=1$	write the $\Sigma=1$ line as your own self-check	technique
$F(x_0)=P(X\le x_0)=\Sigma p(x)$ , $x\le x_0$	note $\le$ , not $<$	MEMORISE
$E(X)=\Sigma x_iP(X{=}x_i)$ ; $\text{Var}(X)=\Sigma x_i^2P(X{=}x_i)-\mu^2$	base definitions	GIVEN
$E(aX{+}b)=aE(X){+}b$ ; $\text{Var}(aX{+}b)=a^2\text{Var}(X)$	square $a$ in the variance line	MEMORISE
Discrete uniform	$p(x)=1/n$ ; no confirmed variance shortcut on this paper — sum $x_i^2p(x_i)$ term by term	technique
$Z=(X-\mu)/\sigma$	standardise first, always	MEMORISE
$\Phi(z)=P(Z<z)$ ; $P(Z>z)=1-\Phi(z)$	deciding when to subtract from 1 is the #1 Normal-dist error (both directions)	GIVEN (table)
Percentage Points table	inverse lookup, $z$ to 4 d.p.	GIVEN (table)
Hypothesis testing ( $H_0/H_1$ , critical value)	does not exist on this paper — confirmed twice in spec guidance, zero hits in the facts bank	N/A

#2. Mark-code cues

Full treatment: forge-exam-technique-guide.md §2.

Code	Meaning	One-line consequence
M	method / attempt at method	M0 → every chained A mark is unreachable
A	accuracy, dependent on the M mark	"M0 A1 is impossible" (verbatim, all 14 schemes)
B	independent, no method needed	often a comment or a graph feature
ft	follow-through	credited even from an earlier wrong value, if the method's right
cao	correct answer only	no partial credit for a near-miss
cso	correct solution only	one upstream slip forfeits it, even if the final line is right
awrt	answers which round to	the one code that relaxes precision
isw	ignore subsequent working	default-forgiving, but overridable per-question (Jan 2024 Q2(c))
oe	or equivalent	covers algebraic AND fraction/decimal/% forms
—	crossed-out work	still marked, unless replaced — never scribble illegibly
—	misread penalty	deduct 2 from A/B marks in the part affected (verbatim, all schemes)

#3. "Conclusion in context" — the four real trigger points

No $H_0/H_1$ on this paper (see §1). WST01's real "context" marks are these four, every one confirmed across multiple series — bare direction/qualitative statements score nothing:

Regression gradient — name the rate, both variables, correct scale. (" $156, not$ 1.56" — Oct 2021 Q2(d); stopped at "positive correlation" — Jun 2022 Q2(d).)
Extrapolation reliability — name the specific value AND the specific range boundary; if no range given, compare to $\bar x$ . "Extrapolation so unreliable" alone is confirmed B0 (Jun 2022 Q2(e) marking note).
Comparing two distributions — name a statistic, give its value for both groups, answer the question actually asked (difference vs similarity). Never claim mean is "more accurate" than median (Jun 2022 Q1(e)).
Modelling assumptions/refinement — name the specific evidence given against an assumption, not a generic true-sounding aside (no standalone past-paper question exists for this point; it rides inside other questions, funded by AO3).

#4. Top-5 danger warnings

Ranked by how independently and repeatedly each is confirmed in the reviewed record — not by invented likelihood. "Seen in N of M series checked" only.

Forgetting to square $a$ in $\text{Var}(aX+b)=a^2\text{Var}(X)$ . Seen in 3 of 6 fully-read examiner-report series, spanning a 3-year gap (Jan 2021 Q4; Jun 2024 Q2(c); Jun 2024 Q3(d)) — the single most repeated numeric error in the whole facts bank.
Conditional probability under a Normal model not recognised at all. The most consistently "most discriminating" sub-skill: flagged in all 5 of the 5 fully-reviewed reports that reach a Normal-distribution conditional question (Jan 2021 Q3(c); Oct 2021 Q6(c); Jun 2022 Q6(c); Jan 2023 Q5(e)).
Subtract-from-1 direction wrong — both ways. Under-subtracting (Jan 2021 Q3(a); Jan 2023 Q5(a)) and over-subtracting (Jun 2024 Q5(a)) both confirmed. Pearson's own fix, stated twice: draw the curve and shade the region before touching the table.
Rounded z instead of the exact 4-d.p. table value. Seen in 3 of 6 series (1.03/1.04 for 1.0364 — Jan 2021 Q3(b); 1.64 for 1.6449 — Oct 2021 Q6(b); 0.25 for the precise equivalent — Jun 2024 Q5(b)).
"Show that" answer with the intermediate step missing. Pearson's own single most-repeated meta-complaint, independent of topic, stated in some form in every one of the 5 general-comment sections read: "many students do not really understand that evidence has to be given" (Oct 2021, general). Confirmed concretely at Oct 2021 Q3(b) (IQR), Oct 2021 Q6(a) (Normal), Jun 2024 Q4(c) (PMCC).

Runners-up worth a glance, each confirmed in 2+ series: independence assumed instead of tested (Jun 2022 general; Oct 2021 Q1(b)); denominator held fixed across without-replacement draws (Jun 2022 Q3(d)); fraction left in a final regression line (Jun 2022 Q2(c)); a blank Venn region not written as 0 (Jun 2022 Q4(c)). Full citations and worked treatment for all of these: forge-exam-technique-guide.md §5 and the individual curriculum lessons it indexes.

#5. Topic weight at a glance

Full table and methodology caveat (6-of-14-series denominator): signal-frequency-and-boundaries.md.

Section 2 (data representation/summary): 6 of 6 checked series — the only topic confirmed on every single one.
Section 6 (Normal distribution): 5 of 6 — and the heaviest single question wherever a full mark breakdown exists (17/75, Jan 2023).
Section 5 (discrete RVs): 5 of 6.
Sections 3 & 4 (probability; correlation/regression): 4 of 6 each.
Section 1 (modelling): 0 of 6 as a standalone question — always embedded in another question.
No grade-boundary data exists in this course's source record for this paper — not stated here because it was never verified, not because it doesn't matter.

#6. Last-minute checklist (10 seconds each)

Full self-mark checklist (30+ lines): forge-exam-technique-guide.md §4. This is the shortest possible version — the five items worth a literal glance right before you turn the page over.

Table values copied to 4 d.p., not rounded.
Every probability fraction simplified — an unreduced $0.15/0.4$ scored A0 for real (Jan 2021 Q1(c)).
Final regression line has no fraction.
$\Sigma p(x)=1$ checked on any discrete probability table.
Any "show that" line has the intermediate step written, not just the target repeated back.

#Sources

research/veridian/WST01-verified-facts.md (primary, provenance-checked facts bank — spec Issue 3 April 2019, formula booklet Issue 2 Jan 2021, 14 series of question papers/mark schemes Jun 2019–Jan 2025, 6 of those series' examiner reports fully read); content/veridian/technique-hub/wst01/ codex-formula-and-method-bank.md, forge-exam-technique-guide.md, and signal-frequency-and-boundaries.md (this paper's already-built consolidated layers, cross-checked for no duplication); the 13 lesson files in content/veridian/curriculum/wst01/. No new PDF fetches, web searches, or primary-source verification were performed for this document — it is a compression pass over an already-verified record, and every specific claim above traces to one of the documents named in this section.