From the Overlap to a Conditional Probability
In Conditional Probability from a Two-Way Table, you found a conditional probability by dividing an overlap count by the total count in the condition group. The same reasoning works when you are given probabilities instead of counts. The formula turns the probability of both events into a probability within the group described by the condition.
For \(P(A\mid B)\), event \(A\) is the target: the event whose probability you want. Event \(B\) is the condition: it defines the group you are considering. The joint probability \(P(A\cap B)\), also written \(P(A\text{ and }B)\), is the probability that both events occur. Divide that overlap probability by \(P(B)\), the probability of being in the condition group.
Each piece has a distinct role. The expression \(P(A\mid B)\) is the answer being sought. The numerator, \(P(A\cap B)\), counts probability from outcomes that satisfy both events. The denominator, \(P(B)\), is the probability of the entire reference group named by the condition. Because that group is the new reference group, the denominator is not automatically 1 and is not automatically the probability of \(A\).
This formula is the probability-based version of the table calculation. If probabilities come from counts in one complete group, the shared grand-total denominator cancels:
That cancellation explains why the formula gives the same result as restricting attention to the condition row or column in a two-way table. With probability inputs, there may be no table to read, but the roles of numerator and denominator stay the same.
A Reliable Way to Use the Formula
Before substituting numbers, translate the question into events. Then read the notation aloud: “probability of \(A\), given \(B\).” This makes it easier to keep the target and condition in their correct places.
In \(P(A\mid B)\), identify the event you want and the event after the bar. The event after the bar defines the reference group.
Use \(P(A\cap B)\) for the numerator and \(P(B)\) for the denominator. Check that both probabilities refer to the same chance process.
The condition must have positive probability: \(P(B)>0\). Also, the overlap cannot be larger than the condition probability.
Calculate the ratio, then describe it as the probability of the target event among outcomes where the condition is true.
A valid conditional probability is between 0 and 1, inclusive. Since the overlap is part of the condition group, \(P(A\cap B)\) cannot exceed \(P(B)\). If a problem gives an overlap probability larger than the condition probability, check whether the values or the event labels have been copied correctly.
The condition probability must also be greater than zero. If \(P(B)=0\), the formula would require division by zero, and there are no outcomes in the stated condition group from which to calculate a proportion. In that case, \(P(A\mid B)\) is undefined.
Worked Example: Reminder Messages for Online Appointments
Worked Example: Reminder Messages for Online Appointments
A fictional appointment system has a model in which 32% of appointments are booked less than 24 hours in advance. Of all appointments, 20% are both booked less than 24 hours in advance and followed by a reminder message. Let \(B\) be the event that an appointment is booked less than 24 hours in advance, and let \(A\) be the event that a reminder message is sent. Find the probability that a reminder is sent, given that the appointment is booked less than 24 hours in advance.
State: The target event is receiving a reminder, \(A\). The condition is booking less than 24 hours in advance, \(B\). The requested probability is \(P(A\mid B)\).
Plan: Use the joint probability \(P(A\cap B)=0.20\) as the numerator and the condition probability \(P(B)=0.32\) as the denominator. The condition probability is positive. The inputs are consistent because the probability of both events, 0.20, is no greater than the probability of \(B\), 0.32.
Do:
The division is \(0.20\div0.32=0.625\). A check using the multiplication relationship gives \(0.625(0.32)=0.20\), which returns the stated joint probability.
Conclude: In this model, the probability that a reminder is sent among appointments booked less than 24 hours in advance is 0.625, or 62.5%.
The numerator is not the probability of reminders in general; it includes only appointments that meet both descriptions. The denominator is not the probability of a reminder; it describes the condition group. Keeping those roles straight is what makes the ratio answer the stated question.
Worked Example: On-Time Deliveries in a Local Order Group
Worked Example: On-Time Deliveries in a Local Order Group
In a fictional set of 300 online orders, 90 are local orders and 72 are both local orders and delivered on time. Let \(L\) mean that an order is local, and let \(T\) mean that it is delivered on time. Find the probability that an order is delivered on time, given that it is local.
State: The target is on-time delivery, \(T\), and the condition is a local order, \(L\). The requested probability is \(P(T\mid L)\).
Plan: The joint count is 72, and the condition-group count is 90. Since both counts come from the same 300 orders, convert them to probabilities using the same grand total, then apply the conditional probability formula. The condition group is not empty, so \(P(L)>0\).
Do:
As a direct count check, restrict attention to the 90 local orders and divide the 72 on-time local orders by that group total: \(72\div90=0.80\). The probability ratio and the count ratio agree.
Conclude: Of the local orders in this fictional set, 80% were delivered on time. Equivalently, the model probability of on-time delivery for an order selected from the local-order group is 0.80.
The count check also shows why dividing the joint count by 300 would answer a different question: \(72/300=0.24\) is the probability that an order is both local and on time. The conditional probability focuses only on local orders, so its denominator is 90, not 300.
Worked Example: Early Arrivals Among Cyclists
Worked Example: Early Arrivals Among Cyclists
A fictional campus transportation model gives \(P(C)=0.25\), where \(C\) means a randomly selected commuter bikes to campus. It also gives \(P(E)=0.30\), where \(E\) means the commuter arrives before 8:30 a.m., and \(P(C\cap E)=0.12\). Find the probability that a commuter bikes to campus, given that the commuter arrives before 8:30 a.m.
State: The target is biking, \(C\), and the condition is arriving before 8:30 a.m., \(E\). The requested probability is \(P(C\mid E)\).
Plan: The overlap probability \(P(C\cap E)=0.12\) goes in the numerator. The probability of the condition, \(P(E)=0.30\), goes in the denominator. The denominator is positive, and the overlap is no greater than the condition probability.
Do:
Check by multiplying the conditional probability by the condition probability: \(0.40(0.30)=0.12\), the given joint probability. The value is also valid because it is between 0 and 1.
Conclude: In this model, 40% of commuters who arrive before 8:30 a.m. bike to campus.
The supplied marginal probability \(P(C)=0.25\) is not needed for this calculation. The formula requires the joint probability for the two events and the probability of the condition. Other information can be useful for a different question, but it does not replace either of those two quantities.
Common Mistakes and AP Exam Tips
- Reversing the denominator. In \(P(A\mid B)\), divide by \(P(B)\), because \(B\) is the condition. Dividing by \(P(A)\) would use a different reference group.
- Using the joint probability as the answer. \(P(A\cap B)\) is the probability that both events occur in the full chance process. The conditional probability is the ratio of that overlap to the condition probability.
- Dividing by a marginal probability for the target. The denominator comes from the event after the bar, not whichever probability seems most familiar or largest.
- Ignoring whether the denominator is positive. The formula applies only when \(P(B)>0\). If the condition has probability zero, the conditional probability is undefined.
- Mixing counts and probabilities. Do not divide a probability by a count, or a count by a probability. Either use counts from the same table and divide overlap count by condition count, or use probabilities and divide \(P(A\cap B)\) by \(P(B)\).
- Giving an interpretation without the reference group. A complete interpretation names the condition group. For example, say “Among appointments booked less than 24 hours in advance,” rather than only “The probability is 0.625.”
A useful check is to multiply your result by the condition probability. The product should equal the joint probability: \(P(A\mid B)P(B)=P(A\cap B)\). This also reinforces that the conditional probability and the joint probability are related quantities, but they answer different questions.
Check Your Understanding
For each question, identify the target event and the condition before calculating.
- In a fictional model, \(P(R\cap S)=0.18\) and \(P(S)=0.45\). Find \(P(R\mid S)\), and explain what the denominator represents.
- A fictional delivery service records 50 orders in a particular category, of which 35 arrive on time. Use counts to find the probability of arriving on time within that category.
- Suppose \(P(M\cap N)=0.14\) and \(P(N)=0.10\). Explain why these values cannot both be correct for the same chance process and event definitions.
- If \(P(A\cap B)=0.24\) and \(P(B)=0\), can \(P(A\mid B)\) be calculated using the formula? Explain.
- A student calculates \(P(A\mid B)\) by dividing \(P(A\cap B)\) by \(P(A)\). Identify the error and state which probability should be the denominator.