Chapter 15

Probability
Detailed Study Notes with Diagrams and Worked Examples
Topics: Successive & Combined Events • Independent Events • Tree Diagrams • Experimental
Probability • Expected Frequency

Introduction to Probability
Probability is the branch of mathematics that measures and predicts how likely events are to occur. It
underpins statistics, gambling, insurance, medicine, weather forecasting, artificial intelligence, and
almost every field where decisions are made under uncertainty. The probability of an event is always a
number between 0 and 1 (or equivalently, between 0% and 100%).

Probability = 0: The event is impossible. It will never happen. E.g. rolling a 7 on a standard die.
Probability = 1: The event is certain. It will always happen. E.g. rolling a number less than 7 on a
standard die.
Probability = 0.5: Even chance. Equally likely to happen or not. E.g. flipping a fair coin and getting
heads.
KEY FACT: Probability can be written as a fraction, decimal, or percentage. They are all
equivalent: 1/4 = 0.25 = 25%.
1. Successive and Combined Events
A combined event is when we consider two or more individual events together, either happening at the
same time (simultaneously) or one after another (successively). We need systematic methods to list all
possible outcomes and calculate probabilities.

1.1 Sample Space Diagrams

A sample space is the complete set of all possible outcomes. A sample space diagram (also called a
possibility space) organises all outcomes from two events into a grid, making it easy to count
favourable outcomes.

Worked Example — Rolling a Die and Flipping a Coin:

Total outcomes = 6 (die) × 2 (coin) = 12 equally likely outcomes.
P(even number) = outcomes with even die / total = 6/12 = 1/2
P(6 and Heads) = 1/12 (only one yellow cell in the grid)
P(odd number or Tails) = count favourable cells: (1,T),(3,T),(5,T),(1,H),(3,H),(5,H),(2,T),(4,T),(6,T)
= 9/12 = 3/4 (be careful not to double-count (1,T),(3,T),(5,T))
KEY FACT: Always count carefully from the grid. Do not rely on mental arithmetic alone when
dealing with combined outcomes.

1.2 Two Dice — A Deeper Example

Rolling two dice produces 6 × 6 = 36 equally likely outcomes. The sum of the two dice ranges from 2 to
12, but the probabilities are NOT all equal — there are more ways to make some sums than others.
Worked Example — Probability of Each Sum:
P(sum = 7) = 6/36 = 1/6 (most likely sum: six ways: 1+6, 2+5, 3+4, 4+3, 5+2, 6+1)
P(sum = 2) = 1/36 (least likely: only 1+1)
P(sum > 9) = P(10)+P(11)+P(12) = 3/36+2/36+1/36 = 6/36 = 1/6
P(sum is prime) = P(2)+P(3)+P(5)+P(7)+P(11) = 1+2+4+6+2 = 15/36 = 5/12
WATCH OUT: The number 7 is the most likely sum, not 6 or 8 — this surprises many people.
The symmetry of the distribution is centred on 7.

1.3 The AND Rule and OR Rule

When we combine events using "AND" or "OR", we apply specific rules to calculate the combined
probability. These rules depend on whether events are mutually exclusive or independent.

The AND Rule (Multiplication Rule)

For two INDEPENDENT events A and B (where the outcome of one does not affect the other):
Example 1:
A bag has red and blue balls. P(Red) = 3/5. A die is rolled. P(6) = 1/6.
 P(Red ball AND roll a 6) = 3/5 × 1/6 = 3/30 = 1/10 [these are independent events]
Example 2 — Three events:
P(three heads in a row) = P(H) × P(H) × P(H) = 1/2 × 1/2 × 1/2 = 1/8
The OR Rule (Addition Rule)
For MUTUALLY EXCLUSIVE events A and B (they cannot both happen at the same time):
Example:
Roll a fair die. P(rolling a 1 or a 6) = P(1) + P(6) = 1/6 + 1/6 = 2/6 = 1/3
These are mutually exclusive because you cannot roll both a 1 AND a 6 simultaneously.
The Complementary Rule
Example:
P(not rolling a 6) = 1 - P(6) = 1 - 1/6 = 5/6
P(at least one head in 4 flips) = 1 - P(no heads) = 1 - (1/2)^4 = 1 - 1/16 = 15/16
KEY FACT: The complement rule is often the most efficient approach for 'at least one'
problems. Finding P(none) is usually much simpler.
2. Independent Events
Two events are INDEPENDENT if the outcome of the first event has absolutely no effect on the
probabilities of the second event. This is a crucial concept — confusing independent with dependent
events is one of the most common mistakes in probability.

2.1 Understanding Independence

Classic Independent Events

•​ Flipping a coin multiple times: each flip is always 1/2 regardless of previous results.
•​ Rolling two dice simultaneously or successively: one die does not influence the other.
•​ Drawing cards WITH replacement: you put the card back, so the deck is always the same.
•​ Drawing from two separate bags: one bag cannot influence the contents of the other.
Classic Dependent Events (NOT Independent)
•​ Drawing cards WITHOUT replacement: the deck changes with each draw.
•​ Drawing balls from a bag without replacement: probabilities change after each draw.
•​ Choosing people for a team without putting them back: the group shrinks.
WATCH OUT: The 'Gambler's Fallacy': after 5 heads in a row, many people think tails is 'due'.
This is WRONG. Each flip is independent. P(heads) is ALWAYS 1/2, no matter what happened
before.

2.2 Testing for Independence

Two events A and B are independent if and only if: P(A and B) = P(A) × P(B). You can check this by
seeing whether the product of the individual probabilities equals the probability of both happening
together.
Example — Testing independence:
In a class, P(likes maths) = 0.6, P(has glasses) = 0.3, P(likes maths AND has glasses) = 0.18.
Check: P(M) × P(G) = 0.6 × 0.3 = 0.18 = P(M and G)
Since the product equals the combined probability, the events ARE independent.
This means wearing glasses has no relationship with liking maths in this class.

2.3 Successive Independent Events

When a series of independent events occur one after another, the probability of any specific sequence
is found by multiplying all the individual probabilities together. The probability of each event remains
unchanged for every trial.
Example 1 — Multiple coin flips:
P(exactly HTHH in 4 flips) = 1/2 × 1/2 × 1/2 × 1/2 = 1/16
Note: EVERY specific sequence of 4 flips has probability 1/16.
Example 2 — Dice rolls:
P(rolling three 6s in a row) = 1/6 × 1/6 × 1/6 = 1/216
P(rolling no 6s in three rolls) = 5/6 × 5/6 × 5/6 = 125/216
P(at least one 6 in three rolls) = 1 - 125/216 = 91/216 ≈ 0.421
KEY FACT: To find P(at least one success), use: 1 - P(all failures). This avoids having to add up
many cases.
3. Tree Diagrams
A tree diagram is a visual representation of all possible outcomes of two or more successive events.
Each branch represents one possible outcome, with the probability of that outcome written on the
branch. Tree diagrams are especially powerful when events are dependent (probabilities change
between draws) or when events have unequal probabilities.

3.1 How to Draw and Use a Tree Diagram

Rules for Tree Diagrams
•​ MULTIPLY probabilities along each path (branch to branch) to find the probability of that specific
sequence.
•​ ADD the probabilities of different paths to find the probability of outcomes that can happen in
more than one way.
•​ The probabilities on branches from the same point must always ADD UP to 1.
•​ The probabilities of ALL final outcomes must sum to 1 (use this to check your work).

3.2 Tree Diagram — With Replacement

When items are replaced before the next draw, probabilities stay the same on every branch at the same
level. The two events are INDEPENDENT.

Full worked solution (bag: 3 Red, 2 Blue — WITH replacement):

P(RR) = 3/5 × 3/5 = 9/25
P(RB) = 3/5 × 2/5 = 6/25
P(BR) = 2/5 × 3/5 = 6/25
P(BB) = 2/5 × 2/5 = 4/25
Check: 9/25 + 6/25 + 6/25 + 4/25 = 25/25 = 1 ✓
P(both same colour) = P(RR) + P(BB) = 9/25 + 4/25 = 13/25
 P(at least one Red) = 1 - P(BB) = 1 - 4/25 = 21/25
P(one of each colour) = P(RB) + P(BR) = 6/25 + 6/25 = 12/25

3.3 Tree Diagram — Without Replacement

When items are NOT replaced, the probabilities on the second set of branches change because the
total number of items in the bag has decreased by 1, and the composition has also changed. The
events are DEPENDENT.

Full worked solution (bag: 3 Red, 2 Blue — WITHOUT replacement):

First draw Red (3/5): now 4 items remain — 2 Red, 2 Blue.
P(RR) = 3/5 × 2/4 = 6/20 = 3/10
P(RB) = 3/5 × 2/4 = 6/20 = 3/10
First draw Blue (2/5): now 4 items remain — 3 Red, 1 Blue.
P(BR) = 2/5 × 3/4 = 6/20 = 3/10
P(BB) = 2/5 × 1/4 = 2/20 = 1/10
Check: 3/10 + 3/10 + 3/10 + 1/10 = 10/10 = 1 ✓
P(same colour) = P(RR) + P(BB) = 3/10 + 1/10 = 4/10 = 2/5
P(different colours) = P(RB) + P(BR) = 3/10 + 3/10 = 6/10 = 3/5
WATCH OUT: Do NOT use the same probabilities on the second set of branches when drawing
without replacement. The denominator decreases by 1 AND the numerator of whichever colour
was drawn also decreases by 1.

3.4 Three-Event Tree Diagrams

Tree diagrams can be extended to three or more events, though they become wider with each
additional stage. The same rules apply: multiply along branches, add across paths for "or" questions.
Example — Three coin flips:
All 8 paths each have probability (1/2)³ = 1/8.
 P(exactly 2 heads) = P(HHT)+P(HTH)+P(THH) = 1/8+1/8+1/8 = 3/8
P(at least 2 heads) = P(HHT)+P(HTH)+P(THH)+P(HHH) = 4/8 = 1/2
Note: from the 8 equally likely outcomes, just count favourable ones.
KEY FACT: For repeated independent events, it is often quicker to use the counting method:
how many arrangements have the required outcome? Then divide by total arrangements.
4. Experimental Probability
Experimental probability (also called relative frequency) is the probability estimated from actually
carrying out an experiment or observing real data. Unlike theoretical probability (which is calculated
from perfect, idealised models), experimental probability is based on what actually happens. The two
approaches should agree if the experiment is repeated enough times — this is one of the most
fundamental ideas in statistics.

4.1 Definition and Formula

This is also called the relative frequency of the event. The more trials are conducted, the more reliable
this estimate becomes. With only a few trials, results can deviate significantly from theoretical values by
chance alone.

4.2 Convergence to Theoretical Probability

The Law of Large Numbers states that as the number of trials increases, the experimental probability
converges towards the theoretical probability. With 10 trials, results can be wildly different. With 10,000
trials, results should be very close to the theoretical value.

Example — Coin flipping experiment:

After 10 flips: 7 heads → experimental P(H) = 7/10 = 0.70 (theoretical: 0.50)
After 50 flips: 28 heads → experimental P(H) = 28/50 = 0.56
After 500 flips: 254 heads → experimental P(H) = 254/500 = 0.508
After 5000 flips: 2491 heads → experimental P(H) = 2491/5000 = 0.498
Conclusion: as trials increase, experimental P(H) gets much closer to 0.5.
WATCH OUT: After getting 5 heads in a row, some people think tails is 'overdue'. This is the
Gambler's Fallacy — each flip is completely independent.

4.3 Comparing Experimental and Theoretical Probability

By comparing experimental results to theoretical values, we can test whether a coin, die, or spinner is
fair (unbiased) or biased. A large difference between experimental and theoretical probability —
especially over many trials — is strong evidence of bias.

Example — Testing a die for bias:

Theoretical P(6) on a fair die = 1/6 ≈ 0.167
Roll a die 300 times. Result: 6 appears 72 times.
Experimental P(6) = 72/300 = 0.24
0.24 >> 0.167: this is strong evidence the die is biased towards 6.
Note: we say "evidence of bias", not "proof of bias" — probability never gives certainty.

4.4 When to Use Experimental Probability

Experimental probability must be used when theoretical probability cannot be calculated — for
example, when outcomes are not equally likely and we do not know the exact probabilities. This is
common in real-world situations:
•​ Weather forecasting: "70% chance of rain tomorrow" is based on historical data, not equal
outcomes.
•​ Medical trials: probability of side effects is estimated from trial data.
•​ Insurance: probability of a car accident is calculated from past accident rates.
•​ Quality control: probability a product is defective is measured from samples.
•​ Sports: probability of a team winning is estimated from past performance data.
KEY FACT: Experimental probability gives us real-world estimates. The larger the sample size,
the more reliable the estimate. Always look critically at how many trials were conducted before
trusting a result.
5. Expected Frequency
If we know the probability of an event and the number of trials we plan to conduct, we can predict how
many times we would EXPECT the event to occur. This is called the expected frequency. It is a
prediction — in reality, the actual frequency may differ slightly due to chance, but should be close if the
number of trials is large.

5.1 The Expected Frequency Formula

This formula works for both theoretical probabilities and experimental probabilities. Expected frequency
is always a predicted value — it is rarely exactly achieved in practice, but gives us a target to compare
actual results against.

5.2 Worked Examples

Example 1 — Simple expected frequency:
A fair die is rolled 300 times. How many 6s are expected?
P(6) = 1/6. Expected frequency = 1/6 × 300 = 50 sixes.
In practice you might get 47 or 54 — both are close to 50 and expected due to chance.
Example 2 — Spinner with unequal sections:
Spinner: Red (1/4), Blue (1/4), Green (3/8), Yellow (1/8). Spun 400 times.
Expected Red: 1/4 × 400 = 100
Expected Blue: 1/4 × 400 = 100
Expected Green: 3/8 × 400 = 150
Expected Yellow: 1/8 × 400 = 50
Total check: 100+100+150+50 = 400 ✓
Example 3 — Using experimental probability for prediction:
In 80 trials, a drawing pin lands point-up 52 times.
Experimental P(point-up) = 52/80 = 0.65
Expected point-up in 500 trials: 0.65 × 500 = 325 times
Example 4 — Comparing expected and actual (testing for bias):
A coin is flipped 200 times. Expected: 100 heads. Actual: 127 heads.
 Experimental P(H) = 127/200 = 0.635, compared to theoretical 0.5.
This is a significant difference — suggests the coin may be biased.
However: we should repeat with more trials to be more confident before concluding bias.

5.3 Expected Frequency vs Actual Frequency

Expected frequency is a theoretical prediction. Actual frequency is what really happens in an
experiment. The gap between them decreases as the number of trials increases. A large persistent gap
— especially over thousands of trials — suggests something is different from our assumed probability
(e.g. a biased die).
Summary comparison:
Expected frequency tells you: "If the probability is p, you would typically get p×n results in n trials."
Actual frequency tells you: "This is what actually happened."
If actual >> expected: event happens more often than theory predicts → possible bias or wrong
probability.
If actual ≈ expected: results consistent with theoretical probability.
KEY FACT: Expected frequency is useful for planning experiments, predicting outcomes in
manufacturing and quality control, and spotting biased or unfair equipment.

WATCH OUT: Expected frequency is NOT guaranteed to be achieved — it is an average

prediction over many repetitions of the experiment. Getting 47 heads in 100 flips (expected: 50)
does not prove a coin is biased.

Chapter 15 — Summary of All Key Formulas and Rules

P(event) = favourable outcomes / total outcomes
P(A') = 1 - P(A) [complement rule]
P(A and B) = P(A) × P(B) [independent events - AND rule]
P(A or B) = P(A) + P(B) [mutually exclusive - OR rule]
P(path on tree) = product of all branch probabilities
Experimental P = frequency / total trials
Expected frequency = probability × number of trials

Independent events: Outcome of one does NOT affect probabilities of the other.
Dependent events: Outcome of one DOES change the probabilities of the other (e.g. without
replacement).
Mutually exclusive: Two events that CANNOT happen at the same time.
Complementary events: Two events where one must happen: P(A) + P(A not) = 1.
KEY FACT: Always check: do all probabilities from the same branch point sum to 1? Do all
final outcomes sum to 1?

— End of Chapter 15 Probability Notes —