Deep Dive: Conditional Probability Independence vs Mutually Exclusive

#CODEX Deep-Dive — Conditional Probability, Independence, and Mutually Exclusive Events (WST01, spec 3.2/3.3)

Provenance (2026-09-29): Read/checked for this document — research/veridian/WST01-verified-facts.md in full, grepped specifically for "independen", "mutually exclusive" and "conditional" (§1 spec transcription, §5 Section-3 trap catalogue, §7 topic-richness ranking, §8/§9 coverage flags, and the two 2026-09-12 per-lesson audit entries for conditional-probability-independence-vs-mutually-exclusive.ts and elementary-probability-and-conditional-notation.ts); the live lesson file content/veridian/curriculum/wst01/conditional-probability-independence-vs-mutually-exclusive.ts in full; content/veridian/technique-hub/wst01/codex-formula-and-method-bank.md, forge-exam-technique-guide.md and signal-frequency-and-boundaries.md (grepped for this topic); content/veridian/question-bank/wst01.ts (grepped — no item on this specific topic exists there yet). No new PDF fetches were made for this document; every specific claim below traces to one of those sources, named inline. All arithmetic in the micro-examples was independently checked with exact-fraction arithmetic (Python's fractions.Fraction) before being written in.

#Scope — what this adds, and what it deliberately does not repeat

The live lesson already derives the conditional probability formula from the multiplication law, proves that mutually exclusive events with positive probability can never be independent, and works two full tree-diagram chains plus a three-event Venn-diagram algebra chain. The formula bank (codex-formula-and-method-bank.md) already tabulates every relevant formula with its memorise/given status. The frequency doc (signal-frequency-and-boundaries.md) already gives this pairing's series-count ranking. None of that is reproduced here.

This document instead does three things the existing material doesn't: (1) states the underlying model formally and proves the piece the lesson asserts but doesn't derive — why the three independence forms are equivalent; (2) covers one real, distinct sub-case the lesson's own trap-taxonomy quotes but never actually computes — conditional probability compounded with a complement event — and, in doing so, reconstructs exactly what the lesson's most-quoted trap number (0.35/0.4) actually is, which neither existing file states explicitly; (3) assembles every real exam angle this specific pairing has been tested from into one place, organised by angle rather than by lesson stage, with the honest limits of that evidence stated plainly.

#1. The formal model

Sample space and mutually exclusive events (spec 3.2). An event is a subset of the sample space $\Omega$. $A$ and $B$ are mutually exclusive when $A \cap B = \varnothing$, i.e. $P(A \cap B) = 0$ — a statement about the sets themselves, fixed before any probabilities are attached.

Conditional probability (spec 3.2). From the multiplication law printed on the WST01 formula sheet, $P(A \cap B) = P(A)P(B\mid A) = P(B)P(A \mid B)$ (the two forms are equal because $A \cap B = B \cap A$). Dividing the second equality by $P(B)$, which must be strictly positive:

$$P(A \mid B) = \frac{P(A \cap B)}{P(B)}.$$

Independence (spec 3.3) — why the three spec forms are one fact, not three. The spec's guidance box lists $P(B\mid A)=P(B)$, $P(A\mid B)=P(A)$ and $P(A\cap B)=P(A)P(B)$ as three separate things to memorise. They are, in fact, algebraically forced to rise or fall together, and it's worth seeing why rather than trusting it: start from $P(A\mid B)=P(A)$ and substitute the conditional-probability formula above, $\frac{P(A\cap B)}{P(B)} = P(A)$, which rearranges to $P(A \cap B) = P(A)P(B)$ — the third form. Running the same substitution the other way round, starting from $P(B \mid A) = \frac{P(A\cap B)}{P(A)}$, reaches the identical product form. So any one of the three, once true, forces the other two — which is exactly why a real mark scheme only ever needs to see one of them tested, not all three (the marked-solution in the live lesson tests only the product form, $P(M)\times P(P)$ vs $P(M\cap P)$, and that alone is a complete, correctly-earning method).

The complete relationship map for two events with $P(A)>0$, $P(B)>0$. There are exactly three possibilities, not two: mutually exclusive (never independent — the lesson's mechanism block proves this: $P(A\cap B)=0$ from one definition and $P(A\cap B)=P(A)P(B)>0$ from the other cannot both hold); independent (never mutually exclusive, by the same proof run the other way); or neither — genuinely overlapping, genuinely dependent, and only settled by running the product-rule calculation. Every one of the four mcq and three prequestion items in the live lesson is really an instance of a student being asked to place a given pair of events into one of these three regions correctly, rather than guessing by eye — which is precisely the chronic error the examiner reports name (§3 below).

#2. Real, distinct sub-cases, each with a short micro-example

#2.1 The independence test and the mutual-exclusivity test are different calculations, run on the same two numbers

A concrete pair from each of the three regions, using a single fair six-sided die (fresh numbers — the live lesson's own examples use two dice, and 5-red/3-blue and 6-red/4-blue counter bags, so none of these three collide with existing content):

• Independent, not mutually exclusive. Standard 52-card deck: $A=$"card is a King", $B=$"card is a Heart". $P(A)=\frac{1}{13}$, $P(B)=\frac14$, $P(A\cap B)=\frac{1}{52}$ (the King of Hearts). $P(A)\times P(B) = \frac{1}{13}\times\frac14=\frac{1}{52}=P(A\cap B)$ — independent, and genuinely overlapping (there is a King of Hearts), consistent with §1's map.
• Mutually exclusive. Die: $A=\{1,2\}$, $C=\{5,6\}$. $A\cap C=\varnothing$ by construction, so mutually exclusive by definition — no calculation needed for that half. Checking independence anyway: $P(A)\times P(C)=\frac13\times\frac13=\frac19\neq 0=P(A\cap C)$, confirming (as the general proof guarantees) that they cannot also be independent.
• Neither. Die: $A=\{1,2,3\}$, $B=\{1,2\}$. $P(A)=\frac12$, $P(B)=\frac13$, $P(A\cap B)=\frac13$ (both 1 and 2 lie in $A$). $P(A)\times P(B)=\frac16\neq\frac13=P(A\cap B)$ — not independent, and $A\cap B\neq\varnothing$ — not mutually exclusive either. This is the large middle ground the live lesson's mechanism block already names (its own dice example, $A=$"first die 6", $B'=$"sum is 8", is a second real instance of it) — included here as a second, shorter illustration with different numbers, not a repeat of that one.

#2.2 Conditional probability compounded with a complement — and what the lesson's own most-quoted trap number actually is

Spec 3.2 lists $P(A')=1-P(A)$ alongside conditional probability, but the live lesson never computes a conditional probability where the conditioning or intersected event is a complement — $P(A\mid B')$ or $P(A\cap B')$ — even though this is exactly the compound skill behind its own most-repeated trap-taxonomy quote. The sibling lesson elementary-probability-and-conditional-notation.ts was independently audited against the real anchor for that quote (Jan 2021 Q1, a three-event Venn diagram: $A$-only$=0.15$, $A\cap B=0.20$, $B$-only$=0.17$, $B\cap C=0.23$, $C$-only$=0.12$, none$=0.13$) and found $P(A\mid B')=P(A\cap B')/P(B') = 0.15/0.4 = \frac38$.

Reconstructing this fully (verified again here by exact arithmetic): $P(A)=0.15+0.20=0.35$ and $P(B')=0.15+0.12+0.13=0.4$. The wrong answer the examiner report records is $0.35/0.4=\frac78$ — which is exactly $P(A)/P(B')$: the correct compound numerator $P(A\cap B')=0.15$ was silently replaced by the plain marginal $P(A)=0.35$, while the denominator $P(B')=0.4$ stayed correct. This matters for how the trap is usually described: "a raw ratio of two marginals with no intersection at all" is not quite the full picture for this specific anchor — the denominator here really is a legitimate conditional-probability denominator ($P(B')$, a genuine complement probability), so the fast self-check "is there an intersection anywhere in this fraction" doesn't by itself catch this version of the error. The mark scheme also requires the fraction simplified: $0.15/0.4$ left unreduced does not earn the accuracy mark, only $\frac38$ or $0.375$ does (verified via the sibling lesson's own audit of this anchor).

Fresh micro-example, same compound skill, different numbers. $P(A)=0.6$, $P(B)=0.45$, $P(A\cap B)=0.3$. Find $P(A\mid B')$. Two routes, both exact: $P(B')=1-0.45=0.55$; $P(A\cap B')=P(A)-P(A\cap B)=0.6-0.3=0.3$ (splitting $A$ into its part inside $B$ and its part outside $B$ — the same identity the real anchor's own mark scheme credits as an alternative route). Then $P(A\mid B')=\frac{0.3}{0.55}=\frac{6}{11}\approx0.545$.

#2.3 Solving for an unknown from the independence product rule directly

The live lesson's own worked-chain solves a two-unknown system where the independence/conditional-probability condition puts the unknown on both sides (Oct 2021 Q1(c) territory) — a genuinely harder version of this skill, not reproduced here. The plain version, one unknown, one equation: $P(A)=0.2$, $A$ and $B$ independent, $P(A\cap B)=0.06$. Since independent, $P(A\cap B)=P(A)\times P(B)$, so $0.06=0.2\times P(B)$, giving $P(B)=0.3$. This is the shape of the M1 in Oct 2021 Q1(c) with the unknown appearing only once rather than on both sides — the entry-level version of the same real skill.

#2.4 Constructing an event to satisfy stated overlap/exclusivity conditions

Oct 2021 Q1(d) asks for an event $D$ drawn on a Venn diagram that is mutually exclusive with a named event $A$ but still overlaps a second named event $B$ — a constructive skill (given two events, build a third meeting stated relations to both) the live lesson covers once, via an MCQ built on its own two-dice scenario. A second, minimal illustration with fresh numbers: spinner with outcomes $1$–$8$, $A=\{1,2,3\}$, $B=\{3,4,5,6\}$. Required: $D$ with $D\cap A=\varnothing$ and $B\cap D\neq\varnothing$. $D=\{4,5\}$ satisfies both: it shares nothing with $A$'s set $\{1,2,3\}$, and it shares $\{4,5\}$ with $B$. The general recipe this exposes: choose $D$ entirely from outcomes outside $A$, then require at least one of those outcomes to also sit inside $B$ — the same two-condition check the real anchor's B1 mark rewards.

#2.5 Flagged extension, not confirmed tested: independence is preserved under taking a complement

This follows in one line from definitions already on the spec (3.2's $P(A')=1-P(A)$, 3.3's product rule) but is not itself listed in the spec's guidance box, and no citation in the reviewed examiner-report archive was found testing it directly — flagged here rather than asserted as a confirmed exam angle. If $A$ and $B$ are independent, then $A$ and $B'$ are independent too: $P(A\cap B')=P(A)-P(A\cap B)=P(A)-P(A)P(B)=P(A)\bigl(1-P(B)\bigr)=P(A)P(B')$ — exactly the product-rule condition, now for $A$ and $B'$. The identical argument applied to $A'$ shows $A'$ and $B$, and $A'$ and $B'$, are independent too. Micro-example: $P(A)=0.5$, $P(B)=0.4$, independent, so $P(A\cap B)=0.2$. Then $P(B')=0.6$, and the theorem predicts $P(A\cap B')=P(A)P(B')=0.3$ — matching the direct check $P(A)-P(A\cap B)=0.5-0.2=0.3$ exactly. Worth knowing as a fallback if a multi-part question ever pairs independence with complement notation, but it should not be presented to a student as a confirmed exam pattern, because it isn't one in the material reviewed.

#3. Synthesis — every real exam angle this pairing has actually been tested from

Confirmed in the archive, one row per distinct angle (not per mark), all traced to WST01-verified-facts.md:

Honest frequency statement, no invented likelihoods. This pairing (independence vs mutual exclusivity, tested by calculation rather than by eye) is directly quoted or evidenced in 4 of the 5 examiner-report series fully reviewed for this facts bank — Jan 2021, Oct 2021, Jun 2022, Jan 2023. The topic-richness ranking in the facts bank's §7 describes it as appearing in all 5, but the fifth series' own citations (Jun 2024) are for conditional probability applied to the Normal distribution — spec point 6.1, a different guidance box, taught in the sibling lesson normal-distribution-standardisation-and-conditional-probability.ts — not a fresh instance of the independence-vs-mutual-exclusivity confusion itself. Both statements can be true at once (conditional probability as a general skill is very widely tested; this exact confusion is confirmed in 4 series), and it's worth keeping them separate rather than rounding one into the other.

#4. Open flags — where the real evidence currently runs out

• Jan 2023 Q2(d)'s "extra probability folded into the numerator" error is confirmed real but not yet drilled anywhere in the live lesson as its own item (it appears only in the trap-taxonomy text, not in a worked-chain, chain-drill or MCQ built around it) — a genuine, named gap, not a claim this document is asked to close.
• 8 of the 14 mark-scheme series that exist for this unit have no confirmed, independently-verified examiner report read yet (Jun 2021, Jan 2022, Oct 2022, Jun 2023, Oct 2023, Jan 2024, Oct 2024, Jan 2025 — per the facts bank's §8/§9). The one spot-check done outside the original five (June 2019) surfaced real gaps in two other topics (outliers/box plots; discrete random variables) and found nothing new for this pairing — but that is one series out of eight unread, and should be read as a lower bound, not evidence this topic's coverage is exhausted.
• §2.5's complement-independence extension is a derivation, not a confirmed exam angle — stated as such above, and it should stay flagged that way rather than be taught as a pattern Pearson has actually set.

#Sources

research/veridian/WST01-verified-facts.md — §1 (spec text), §5 Section 3 (all examiner-report quotes cited above), §7 (topic-richness ranking), §8/§9 (coverage limits), and the two 2026-09-12 audit entries for conditional-probability-independence-vs-mutually-exclusive.ts and elementary-probability-and-conditional-notation.ts. content/veridian/curriculum/wst01/conditional-probability-independence-vs-mutually-exclusive.ts (live lesson, read in full). content/veridian/technique-hub/wst01/codex-formula-and-method-bank.md, forge-exam-technique-guide.md, signal-frequency-and-boundaries.md. content/veridian/question-bank/wst01.ts (checked, no overlapping item found).