A Toronto casino banned a roulette player who bet $200,000 on red after observing 9 consecutive black results. The casino wasn’t worried about his luck — they were worried about his reasoning. The player believed that after 9 blacks, red was “due.” But the wheel has no memory.
This is called the gambler’s fallacy — the belief that past random outcomes influence future independent ones. It’s one of the most pervasive errors in human reasoning, and it costs people millions of dollars every year in casinos, stock markets, and lottery ticket sales.
The mathematics that exposes the gambler’s fallacy — and replaces superstition with rigorous reasoning — is probability. This lesson is where that reasoning begins.
Understanding exactly why the roulette player was wrong requires a concept called independence — the formal idea that one spin cannot influence the next. That definition arrives in PR-2. But independence is built on foundations you need first: sample spaces, events, and the rules that govern how probabilities combine. Without these, the conversation about independence has no language to stand on.
After this lesson, you will be able to:
Identify the sample space of an experiment and express events as subsets of it.
Compute classical, empirical, and subjective probabilities and recognize which type applies in context.
Apply the complement rule and the General Addition Rule to probabilities of compound events.
Identify mutually exclusive events using Venn diagrams and distinguish them from independent events.
This is Lesson PR-1. Everything that follows in this module — conditional probability, random variables, distributions — is built on the concepts in this lesson.
Section 2: Prerequisites
▾
What you need from DS-1
Reading two-way frequency tables (DS-1, Sampling): PR-1 uses two-way tables as its primary data format. You need to read cell counts, row totals, and column totals accurately — today we convert those counts to probabilities.
Relative frequency as proportion (DS-1, Data Vocabulary): You already know that a relative frequency is a count divided by the total. Empirical probability is exactly this ratio applied to event outcomes.
Qualitative variables (DS-1, Variable Types): Most probability examples involve categorical outcomes (heads/tails, pass/fail, product A/B). Being fluent with categorical data makes the event language natural.
Retrieval Checkpoint
A school surveyed 120 students about their study method and exam result. The two-way table below shows the results.
Passed
Failed
Total
Studied in groups
48
12
60
Studied alone
42
18
60
Total
90
30
120
What is the relative frequency of students who studied in groups and passed?
Also confirm you can recall:
Success Factor:
From description to prediction. In DS-1, relative frequencies described data that had already been collected — they summarized what had happened. In PR-1, we treat those frequencies as probabilities — numbers that predict what is likely to happen before we see the outcome. The arithmetic is identical; the interpretation is new.
Retrieval Warm-up — from earlier lessons
A data set has mean and standard deviation . A value of lies how many standard deviations above the mean?
A symmetric, approximately normal data set has mean 100 and standard deviation 15. According to the Empirical Rule, about what percentage of values fall between 70 and 130?
Section 3: Core Concepts
▾
Navigation Guide — 10 Concepts
C1–C4: What probability means — four ways to assign numbers to uncertain outcomes.
C5–C6: The rules every probability must obey, and the most useful shortcut: the complement.
C7–C9: Combining two events with “or” and “and,” culminating in the General Addition Rule.
C10: A critical distinction that trips up students all semester: mutually exclusive ≠ independent.
C1 — Sample Space and Events
Every probability problem begins with an experiment — any process whose outcome is uncertain. We write for the sample space: the set of all possible outcomes. We write to say that event is a subset of — a collection of outcomes that interest us.
Sample Space and Event
The sample space is the set of all possible outcomes of an experiment.
An event is any subset of : .
Event occurs when the actual outcome belongs to .
Example. Roll a fair six-sided die. The sample space is . Let = “roll an even number” . If the die shows 4, event occurs.
C2 — Classical Probability
When every outcome in is equally likely, probability is a simple counting ratio.
We write for the number of outcomes in event .
Classical Probability
For an experiment with equally likely outcomes:
Example. With the die above: , , so .
Classical probability requires equally likely outcomes. If you flip a bent coin, the sample space is still , but — the formula gives the wrong answer for non-uniform spaces. A particularly tempting error: “It will rain tomorrow or it won’t — so there’s a 50% chance of rain.” That reasoning applies to a space where outcomes are not equally likely. Rain tomorrow is an empirical or subjective probability, not a classical one.
C3 — Empirical Probability
When we cannot assume equally likely outcomes, we estimate probability from observed data.
Empirical (Relative Frequency) Probability
After trials, the empirical probability of event is:
where is the number of trials in which occurred. As grows, this estimate converges to the true probability (Law of Large Numbers — conceptual).
Example. A factory inspects 400 items and finds 20 defective. The empirical probability of a defective item is .
Try it yourself. The top chart shows the running proportion converging toward 0.50 as grows. The bottom chart shows the gap between the number of heads and tails — and it does not shrink; if anything it drifts upward, tracking the dashed curve. That contrast is the whole point: a run of tails (watch the recent-flips strip) never makes the line “correct” upward. And no matter how long the current streak gets, the predictor below it keeps reading — the coin has no memory. Convergence comes from averaging, not from the counts catching up.
Recent flips:
Click a button to start flipping.
C4 — Subjective Probability
When no repeatable experiment exists and outcomes are not equally likely, probability can express a personal degree of belief.
Enrichment — beyond the assessed standards. Subjective probability — and classifying a probability as classical vs. empirical vs. subjective — is here for context: it sharpens how you read probability claims in the world. The graded standards (PR-1) assess computing classical and empirical probabilities and applying the rules, not naming which type a number is. Read this to think clearly; it won’t appear on a standards assessment.
Subjective Probability
A subjective probability is a numerical degree of belief that an event will occur, assigned by a person based on their knowledge and judgment. It must still satisfy and (C5).
Example. A venture capitalist says: “I estimate a 30% probability this start-up will reach profitability within two years.” No repeatable experiment produces this number — it is an informed subjective judgment.
All three types of probability — classical, empirical, subjective — must obey the same axioms (C5). A subjective probability that “feels right” but violates is internally inconsistent and must be revised.
For each scenario below, decide which type of probability it is before checking — then pick a card to see if you were right.
Pick the type for each scenario below.
Read the scenario, then pick the type of probability it is:
C5 — Probability Axioms
All valid probabilities, regardless of how they are computed, must satisfy three foundational rules.
Probability Rules
For any event and sample space :
— every probability lies between 0 and 1 inclusive.
— the complete sample space has probability 1; something must happen.
— the impossible event has probability 0.
Key consequence: If and are mutually exclusive (), then .
For the curious: the formal Kolmogorov axioms (1933) treat countable additivity — the Key consequence above — as the third axiom, from which is then derived rather than stated separately. This simplified presentation captures everything needed for this course.
Validity check. If someone claims and , the sum is , violating Axiom 2. Valid complementary events must sum to exactly 1.
Every valid probability lives in the interval — but is not a probability. Drag and below: their raw sum (the upper needle) can shoot past 1 into the red zone, and that breaks no axiom — it simply means the sum has counted the overlap twice. Now drag : subtracting it once pulls the lower needle, , back into the valid interval, where a real probability always lives. That subtraction is exactly the General Addition Rule (C8).
not a probability
0Impossible0.5Fair coin1Certain
0.60
0.50
0.30
C6 — Complement Rule
The complement of event consists of all outcomes in not in . Because and together cover all of and never overlap, their probabilities must sum to 1.
Complement Rule
Strategy: When computing directly is hard, compute and subtract from 1. This is especially powerful for “at least one” problems.
Example., so .
Checkpoint — C5 and C6
Question A. Apply the complement rule to find the missing probability.
Question B. A classmate says: ” and too — both outcomes are possible, so both can have the same probability.” What is the problem with this claim?
C7 — Compound Events: Union and Intersection
Two events can be combined using set operations. Understanding these operations precisely prevents the most common errors with the Addition Rule.
Union and Intersection
— the union of and : the event that occurs, or occurs, or both. (At least one.)
— the intersection of and : the event that both and occur simultaneously.
Example. Roll a die. Let (even) and (greater than 3).
— even or greater than 3 (or both).
— even and greater than 3 simultaneously.
Venn diagram: draw two overlapping circles inside rectangle . The left-only region is only; the right-only region is only; the overlap is ; outside both circles is neither.
Click any cell in the table below to see how each table entry maps onto the four Venn regions. This connection bridges the DS-1 frequency table format you already know with the event language used in all probability calculations. Use “Show as probabilities” to flip every count to its proportion (count ÷ 120) — the same diagram, now reading where it read .
Survey of 120 students — click any cell
Passed B
Failed B′
Total
Groups A
48
12
60
Alone A′
42
18
60
Total
90
30
120
Venn diagram (n = 120)
Click any highlighted cell or total to see the corresponding Venn region.
C8 — General Addition Rule
When you add , every outcome in gets counted twice — once as part of and once as part of . The General Addition Rule corrects for this double-counting.
General Addition Rule
For any two events and :
The intersection is subtracted once to undo the double-counting of outcomes that belong to both events.
Example. In a class of 40 students: 18 play a sport, 14 are in a club, and 6 do both.
Without subtracting: — wrong, because the 6 students in both groups were counted twice.
Never add the intersection — always subtract it. A frequent error is . Adding the intersection makes the double-counting worse, not better.
You cannot compute by multiplying unless you know and are independent. The formula requires independence — a condition formally defined in PR-2. In PR-1, always read directly from the data or the problem statement.
Try dragging the sliders below. Each dot is one of 100 students ( = plays a sport, = is in a club) — use “Show as sets” to fade from that concrete picture to the abstract , notation once the idea is clear. Notice that is always less than when the events overlap — because the intersection gets counted twice without the correction. Click “Make ME” to see what changes when the events cannot both happen.
Each dot is 1 of 100 students. A = plays a sport, B = is in a club.
sport onlybothclub onlyneither
30
20
10
C9 — Mutually Exclusive Events
Two events are mutually exclusive (ME) if they cannot both occur in a single trial — their intersection is empty.
Mutually Exclusive Events
Events and are mutually exclusive if (equivalently, ).
For mutually exclusive events, the General Addition Rule simplifies to:
Example. Roll a die. Let = “roll a 1” and = “roll a 6.” One trial cannot show both, so . Therefore .
Always verify mutual exclusivity before using the simplified Addition Rule. Mutual exclusivity is a mathematical condition — — not a semantic one. Two events about “different things” can still share outcomes. Example: “owns a car” and “uses transit” seem unrelated, but one commuter can do both. If you apply to overlapping events, the result may exceed 1, violating the axioms.
C10 — Mutually Exclusive vs. Independent (Preview)
These two terms describe completely different relationships. Mixing them up is one of the most persistent errors in this course.
Mutually exclusive is about overlap in the sample space: — the events cannot co-occur.
Independent is about informational influence: knowing occurred tells you nothing new about . Formally: . This is defined and tested in PR-2.
ME vs. Independence — Critical Distinction
Mutually exclusive: — the events cannot happen at the same time.
Independent (preview): — knowing occurred does not change the probability of .
Two events with positive probability that are mutually exclusive are always dependent. If occurred, definitely did not — so knowing changes the probability of from to 0. That is the opposite of independence.
Notation boundary. We say and are independent when knowing occurred does not change the probability of . This is a precise mathematical condition — not the same as “they seem unrelated.” We will define and test it formally in PR-2. For now: do not use “mutually exclusive” and “independent” as synonyms.
“These events are mutually exclusive, so they must be independent — they can’t affect each other.” This is backwards. If and are mutually exclusive with and , knowing occurred tells you with certainty that did not. That is maximum dependence, not independence.
Click “A occurred →” in each panel to see how the two concepts differ — one rules out entirely, the other leaves completely unaffected.
Mutually Exclusive
Die roll — A = {1, 2}, B = {4, 5, 6}
Independent
Two coins — A = Flip 1 = H, B = Flip 2 = H
Section 4: Worked Examples
▾
Example 1 — Fully Worked: Classical Probability with a Standard Deck
Problem. A card is drawn at random from a standard 52-card deck. Let = “draw a heart” and = “draw a face card (J, Q, K of any suit).” Compute , , , , and .
Show Solution
Metacognitive narration.
I notice that the deck is fair — all 52 cards are equally likely. Classical probability applies.
I need to count the outcomes in each event carefully before computing any probability.
.
(all 13 hearts). So .
(J, Q, K in each of 4 suits). So .
(J♥, Q♥, K♥). So .
I notice that and overlap — the intersection is non-empty. I must use the General Addition Rule, not the simplified one.
For the complement:. This is the probability of drawing a non-face card — the complement shortcut is faster than counting 40 cards directly.
Example 2 — Prediction Checkpoint: Empirical Probability from a Two-Way Table
Problem. A university surveyed 200 students about streaming subscriptions. Results:
Has Netflix
No Netflix
Total
Has Spotify
60
36
96
No Spotify
70
34
104
Total
130
70
200
Let = “has Netflix” and = “has Spotify.”
From the table:
Before computing — make a prediction. Using only and , what do you think equals?
Show Solution
Apply the General Addition Rule:
Check: If you predicted , you see exactly why subtracting the intersection matters — the 60 students with both subscriptions were counted in both the 130 (Netflix column total) and the 96 (Spotify row total).
Example 3 — Details/Summary: Complement + General Addition Rule
Problem. In a group of 500 employees: , , . Find (a) and (b) .
Try Part (b) first. Think about what “neither” means in terms of the complement rule before reading the solution.
Show Solution
Part (a). Apply the General Addition Rule:
Part (b). “Neither remote nor flexible” is the complement of “remote or flexible”:
Pattern to remember:, and . This two-step pattern appears constantly in probability problems.
Section 5: Guided Practice
▾
Problem 1 — Classical Probability and Complement (C1, C2)
Problem 2 — Applying the Complement Rule (C5, C6)
Problem 3 — General Addition Rule (C7, C8)
The following problem gives you , , and in a realistic context. Apply the General Addition Rule to find . Note that in every variant — the events always overlap, so the simplified rule does not apply.
Problem 4 — Venn Diagram and Mutual Exclusivity (C7, C8, C9, C10)
A group of 80 students is surveyed: 48 play a team sport, 35 play a musical instrument, and 12 do both.
Work through the first two computations one step at a time:
Part (c). Are “plays sport” and “plays instrument” mutually exclusive? Justify with a probability.
Part (d). A classmate says: “Since playing sport and playing instrument are two completely different activities, they must be independent.” Choose the best response.
Section 6: Independent Practice
▾
Problem 1 — Empirical Probability from a Two-Way Table (C1, C3)
Problem 2 — Addition Rule from a Two-Way Table (C7, C8, C3)
Problem 3 — Complement Strategy: At-Least-One Problems (C5, C6)
These problems involve multiple independent trials. The problem statement notes that trials are independent — the formal definition arrives in PR-2, but you can use it intuitively here: the outcome of one trial does not affect the others. Your task is to recognize why the complement strategy is more efficient than listing every case directly.
Problem 4 — Find the Error: Probability Reasoning (C8, C9, C10)
Problem 5 — Multi-Step Synthesis (C1–C10)
A class of 120 students is categorized by two attributes: whether they submitted all assignments (yes: 80, no: 40) and whether they passed the final exam (yes: 90, no: 30). Among those who submitted all assignments, 75 passed the final.
Work through the table completion and the two probability computations one step at a time:
Part (d). Are “submitting all assignments” and “passing the final” mutually exclusive? Justify.
Part (e). A classmate says: “Submitting assignments and passing are different things, so they must be independent.” Evaluate this claim.
Mixed Review — Retrieval from Earlier Lessons
These problems draw on concepts from earlier in the course. Attempting them without re-reading prior lessons is the point — retrieval practice strengthens long-term memory more than re-reading.
Review Problem 1 — Standard deviation and IQR
A quality inspector records the fill volumes (mL) for a sample of 7 bottles: 498, 501, 497, 503, 499, 502, 500. Compute the sample standard deviation and identify the IQR. Is any value an outlier by the 1.5×IQR fence rule?
Show Solution
Step 1 — Mean: mL.
Step 2 — Sample variance: Compute each squared deviation:
Step 3 — SD: mL.
Step 4 — IQR: Sort the values: 497, 498, 499, 500, 501, 502, 503. The median is 500. Q1 is the median of the lower half 499 = 498. Q3 is the median of the upper half 503 = 502. So IQR = 502 − 498 = 4.
Step 5 — Outlier fences: Lower fence = Q1 − 1.5 × IQR = 498 − 6 = 492. Upper fence = Q3 + 1.5 × IQR = 502 + 6 = 508. All seven values fall within [492, 508], so no outliers.
Review Problem 2 — Applying the Empirical Rule
A large university records final exam scores that are approximately normally distributed with mean and standard deviation .
(a) Between what two scores do the middle 95% of students fall?
(b) A student scored 54. Express this as a z-score and describe the student’s standing relative to the class.
Show Solution
(a) The Empirical Rule states that approximately 95% of values in a normal distribution fall within standard deviations of the mean. So:
The middle 95% of students scored between 54 and 90.
(b).
A z-score of means the student scored exactly 2 standard deviations below the mean. From the Empirical Rule, about 95% of students scored between 54 and 90, so only about 2.5% of students scored at or below 54. This student is at the very low end of the distribution.
Section 7: Mastery Check
▾
No hints. No scaffolds. These questions test whether you can recall and apply what you have learned without support.
Question 1 — Feynman Test
Explain to a classmate who missed today’s lesson why we cannot always compute . Use a concrete example — numbers or a deck of cards — where this formula gives the wrong answer, and show what the correct calculation is.
0 / 500
Model Answer
Imagine a class of 30 students: 12 play piano, 10 play guitar, and 4 play both. If you add and divide by 30, you get 22/30 — but those 4 who play both got counted twice: once in “piano” and once in “guitar.” The true count of students who play at least one instrument is , giving . The General Addition Rule fixes the double-counting by subtracting the intersection once. The simple formula is only valid when the events are mutually exclusive — when no outcome can belong to both at the same time.
Question 2 — Apply
Question 3 — Error Analysis
A student writes: ” and . Since these events seem unrelated, I will apply the simple Addition Rule: .”
Student’s conclusion: ” — this means it is very likely that at least one of the two events occurs.”
Identify the error in the student’s work.
Show Solution and Correction
Error type: Applied the simplified Addition Rule without verifying that .
Why the result is impossible: violates Axiom 1. This is the clearest possible signal that the simplified rule was applied incorrectly.
Correction: Without knowing , we cannot compute exactly. We can establish bounds: and . The actual answer requires from the data.
Question 4 — Mutually Exclusive vs. Independent
Two events and each have positive probability and are mutually exclusive — they cannot both occur on the same trial. What can you conclude about whether and are independent?
Show Solution
Key distinction:mutually exclusive means (they cannot co-occur); independent means (one tells you nothing about the other). These are different conditions — and for events with positive probability they are incompatible.
If and are mutually exclusive with and , then but , so — the events are dependent. Intuitively: learning that happened tells you did not, which is a large change in information. Independence is defined and tested formally in PR-2.
Self-Assessment
How confident are you with the material in this lesson?
Still confusedReady for the Boss Fight
Section 8: Boss Fight
▾
Choose your path:
📊 Path A — The Pollster
A citywide survey lands on your desk with a colleague’s confident verdict attached. Is the colleague right?
🎙️ Path B — The Referee
A commentator does probability math live on air and arrives at 115%. Something’s off — find it.
📊 Path A — The Pollster
A citywide survey of 500 residents crossed gym membership with healthy-eating habits. A colleague skims the numbers and announces to the room:
“Gym members and healthy eaters are basically separate crowds — one group is about exercise, the other about diet. They barely overlap.”
Here is the full table:
Healthy Eating
Not Healthy Eating
Total
Gym Member
110
90
200
Not a Member
130
170
300
Total
240
260
500
Your job: Decide whether the colleague is right, and back it with numbers. Then say how likely a randomly chosen resident is a gym member, a healthy eater, or both — and how likely they are neither. Interpret what you find for the 500 residents.
Need a way in?
What would have to be true of the table for the colleague’s “they barely overlap” claim to hold?
Which single cell settles it?
Once you have the pieces, how do you count “gym member or healthy eater” without counting the overlap twice — and how do you get “neither” from that?
0 / 500
Show Solution
Task 1.
— this uses the intersection cell, not the row or column total.
Task 2.
Interpretation: 66% of the 500 residents are gym members, or eat healthily, or both.
Task 3..
Verify: “Not a member AND not healthy eating” cell = 170/500 = 0.34. ✓
Task 4.
Mutual exclusivity requires . But . The events are not mutually exclusive — 110 residents are both gym members and healthy eaters. The colleague’s semantic reasoning (“separate things”) does not substitute for the formal mathematical check.
🎙️ Path B — The Referee
A sports commentator says on live television:
“The home team wins 60% of the time and scores first 55% of the time — so their chance of winning OR scoring first is 115%! They are basically unstoppable!”
Your job: Something is wrong — pin down exactly what, and say what the number should be (or what you would need to know to get it).
Need a way in?
What is the fastest tell that 115% is wrong — before you check any method?
What rule did the commentator use, and what does it quietly assume?
What single quantity is missing? If the team wins and scores first in 40% of games, what is the real answer?
0 / 500
Show Solution
Task 1. The commentator applied the simplified Addition Rule: . This rule requires — that winning and scoring first cannot happen in the same game. That condition is not established and is almost certainly false.
Task 2. violates Axiom 1: all probabilities must lie between 0 and 1. The error is applying the simplified Addition Rule to events that overlap — many games where the team wins, they also score first.
Task 3. The missing piece is — the probability of winning AND scoring first in the same game.
With : . The correct probability is 75%, not 115%.
Task 4. “Measuring different things” does not imply mutual exclusivity. A single game can result in both a win and a first score for the home team. Mutual exclusivity requires that both events literally cannot occur in the same trial — which is clearly not the case here. The correct bound is ; the actual value requires the intersection probability.
Section 9: Challenge Problems
▾
Ready for more? These go beyond the lesson objectives.
Challenge 1 — Derive the General Addition Rule from Counting (C7, C8, C9)
The General Addition Rule is . Don’t take it on faith — build it from counting.
For a sample space of equally likely outcomes: start by counting the outcomes in in terms of , , and , and explain why one term must be subtracted rather than added. Then turn your count into probabilities by dividing through by . Finally, check what your formula becomes when and are mutually exclusive.
Show Solution
Count the union. Add and and every outcome that lives in both and has been counted twice — once inside each set. Subtract that overlap exactly once to fix the double-count:
Why subtract, not add? Adding would count those shared outcomes three times. And it is not — multiplication counts ordered pairs of sequential choices, not the outcomes of a single draw.
Divide by . Every count becomes a probability:
The mutually exclusive case. If and cannot both happen, no outcome is shared, so and the subtracted term vanishes: . The simplified rule is not a different rule — it is the general rule with a zero overlap.
Challenge 2 — The Factory Line
Challenge 3 — Non-Uniform Probability Space (C1, C2, C5, C8)
A biased die has . The remaining five faces (1, 2, 3, 4, 5) are equally likely.
Challenge 4 — A Hidden Simplification
Some unions are easier than they look. Read the setup carefully before you reach for a formula — how and relate is the whole puzzle.
Section 10: Solutions Reference
▾
Complete, step-by-step solutions for all problems in Sections 5–9 are available on the solutions page.
If you are stuck: Re-read the Core Concept in Section 3 that applies, then find the Worked Example in Section 4 that maps to it. The solutions page shows reasoning behind every step, not just the final answer.
Quick-Reference Formulas
Probability Axioms:
Complement Rule:
General Addition Rule:
Mutually Exclusive Events (when ):
Mutually Exclusive vs. Independent:
ME: — overlap is impossible
Independent (PR-2): — no informational influence
ME events with positive probability are always dependent
Back to the roulette wheel. You now have the tools to understand exactly why the casino player’s reasoning failed. Each spin is an independent event — knowing the last 9 results gives no information about the next spin. That is a precise mathematical condition (), formally defined in PR-2. The player’s error was treating nine consecutive blacks as evidence that red was “due,” as if the spins were somehow dependent. With the probability rules from this lesson and the independence definition from PR-2, you will be able to diagnose that error precisely.