Choose the Rule Before You Calculate
Probability mistakes often begin with a familiar-looking phrase. A question asks for “A or B,” and a student multiplies the probabilities because multiplication has appeared in earlier problems. Or two events sound unrelated, so the student assumes independence without checking. The calculation may look neat, but it answers a different question or relies on an unsupported model.
In Union Probabilities with Dice Experiments and Reading a Probability Table for “Or” Questions, you learned that inclusive “A or B” means event \(A\) occurs, event \(B\) occurs, or both occur. The union includes the overlap. In The Multiplication Rule for Independent Events, you learned that multiplying \(P(A)\) and \(P(B)\) finds \(P(A\cap B)\) only when \(A\) and \(B\) are independent. These rules answer different questions: one concerns “or,” and the other concerns “and.”
A useful habit is to translate the question before choosing a rule. “Both” points to the intersection \(A\cap B\). Inclusive “or” points to the union \(A\cup B\). Then ask what is known about the overlap. Is it given directly? Can it be found using a conditional probability and the general multiplication rule, as in Unions with Dependent Events? Or is independence stated or supported by the chance process? If none of these applies, the exact union may not be determined by the information available.
A Quick Error-Check Routine
Before using a formula, pause to identify the event being requested and the evidence for any independence assumption. This short routine helps catch several common errors before they reach the arithmetic.
“And” means both events occur; “or” means at least one event occurs. Unless a question says otherwise, “or” is inclusive.
A union counts the overlap as part of both events. Account for that shared region once, not twice.
Use \(P(A)P(B)\) for the overlap only if independence is stated, is justified by the design, or is supported by probability information.
If the overlap is unknown and independence is not established, do not invent a value for the union. Explain what additional information would be needed.
Independence is not a shortcut for “the events have different names,” “the events involve different things,” or “I do not see an obvious connection.” As discussed in Independence and Real-World Assumptions, it is an assumption about how a chance process works. A stated model can specify independence; a process such as separate, unaffected random devices can provide a reason for it; and probability data can be checked using the criteria from Testing Independence with the Product Rule. Without such a basis, do not silently replace an unknown overlap with \(P(A)P(B)\).
Worked Examples
Worked Example: Urgent Messages or Messages with Attachments
In a hypothetical email model, a randomly selected message has probability \(0.32\) of being flagged urgent and probability \(0.28\) of containing an attachment. The probability that a message is both urgent and has an attachment is \(0.12\). What is the probability that it is urgent or has an attachment? A student suggests multiplying \(0.32\) by \(0.28\).
State. Let \(A\) be the event that the message is flagged urgent, and let \(B\) be the event that it contains an attachment. We want \(P(A\cup B)\), not the probability of both events.
Plan. The question uses inclusive “or,” so we will use the general addition rule and subtract the known overlap. The suggested product would represent the overlap only if the events were independent; no such assumption is given. The supplied joint probability lets us find the union without assuming independence.
Do. Substitute the three probabilities into the addition rule:
The sum \(0.32+0.28=0.60\) counts messages in the overlap twice: once among urgent messages and once among messages with attachments. Subtracting \(0.12\) removes the extra count, leaving \(0.48\). The suggested multiplication gives \((0.32)(0.28)=0.0896\), but that is not the requested union. It is the overlap value that would be required by independence, and it differs from the given overlap of \(0.12\).
Conclude. In this hypothetical model, the probability that a randomly selected message is urgent or has an attachment is \(0.48\). Multiplying the two marginal probabilities does not answer the “or” question; the overlap must be handled using the information actually given.
Worked Example: Grocery Orders and an Unsupported Independence Assumption
A hypothetical grocery-order model gives \(P(A)=0.40\) for an order paid for with a digital wallet and \(P(B)=0.30\) for an order scheduled for evening delivery. First suppose these are the only probabilities provided. Can you determine the probability that an order was paid for with a digital wallet or scheduled for evening delivery? Now suppose the model also gives \(P(A\cap B)=0.18\). Find the union and assess the independence assumption.
State. Let \(A\) mean an order was paid for with a digital wallet and \(B\) mean it was scheduled for evening delivery. The target is \(P(A\cup B)\).
Plan. With only \(P(A)\) and \(P(B)\), the general addition rule still needs \(P(A\cap B)\). We must not assume that the overlap equals \(P(A)P(B)\) merely because the two event descriptions are different. When the joint probability is supplied, we can calculate the union directly and compare the actual overlap with the product-rule value.
Do. With only the marginal probabilities, the calculation would have the form \(0.40+0.30-P(A\cap B)\). Since the overlap is not given and independence has not been established, the exact union cannot be determined from those two numbers alone.
When the model supplies \(P(A\cap B)=0.18\), the union is
To check whether independence fits these probabilities, calculate the product of the marginals:
The product is \(0.12\), whereas the stated probability of both events is \(0.18\). Because the values are not equal, these events are not independent in this model. If someone had assumed independence, they would have calculated the overlap as \(0.12\) and obtained \(0.40+0.30-0.12=0.58\), rather than the correct union of \(0.52\).
Conclude. The probability of a digital-wallet payment or evening delivery is \(0.52\) when the model gives an overlap of \(0.18\). The information with only the two marginal probabilities is insufficient for an exact union, and the additional joint probability shows why assuming independence would produce a different answer.
Worked Example: Disjoint Events Are Not Automatically Independent
In a hypothetical volunteer survey, one preferred work shift is recorded for each volunteer. For a randomly selected volunteer, let \(A\) mean the preferred shift is early and \(B\) mean it is late. Suppose \(P(A)=0.25\) and \(P(B)=0.35\). What is the probability that the selected volunteer prefers an early or a late shift? Would it be correct to treat the events as independent?
State. We want the union of the early-shift and late-shift events. Since each volunteer records just one preferred shift, a volunteer cannot prefer both of these recorded categories at once.
Plan. The events are disjoint, so their overlap is zero and the general addition rule simplifies to the sum of the two probabilities. To check independence, compare the zero overlap with the product of the positive marginal probabilities. Disjointness does not justify using the independent-events union formula.
Do. Because \(A\cap B\) is impossible, \(P(A\cap B)=0\). Therefore,
If the events were independent, their overlap would have to be
But the actual overlap is \(0\), not \(0.0875\). The conditional comparison makes the same point: \(P(B\mid A)=0\), because an early-shift preference rules out a late-shift preference, while \(P(B)=0.35\). The condition is valid because \(P(A)=0.25>0\).
Conclude. The probability that a randomly selected volunteer prefers an early or late shift is \(0.60\). The events are not independent: knowing that the volunteer prefers an early shift changes the probability of a late-shift preference from \(0.35\) to \(0\). Treating disjoint events as independent would give an incorrect overlap and union.
Common Mistakes and AP Exam Tips
- Multiplying for “or.” \(P(A)P(B)\) is not a general rule for \(P(A\cup B)\). It can supply the overlap \(P(A\cap B)\) when independence is justified; the addition rule then uses that overlap in the union calculation.
- Adding without subtracting overlap. \(P(A)+P(B)\) counts outcomes in both events twice. Subtract \(P(A\cap B)\), unless the events are disjoint and the overlap is zero.
- Assuming independence from the wording. Events that sound separate are not automatically independent. State the process-based reason, use an explicitly specified model assumption, or check the probability criterion with supplied information.
- Using a product for the overlap without checking. If the events are dependent, \(P(A\cap B)\) may differ from \(P(A)P(B)\). Use the joint probability if it is given, or find the overlap by an appropriate conditional probability method.
- Confusing disjoint with independent. Disjoint events cannot happen together. If both have positive probability, their actual overlap is zero but the product of their probabilities is positive, so they are dependent.
- Claiming an exact union when information is missing. If the overlap is unknown and independence is unsupported, say the union cannot be determined from the information given. Do not fill the gap by treating the events as independent.
On an AP response, a full-credit explanation names the requested event, identifies the rule, and shows why the overlap value is valid. For example: “The events are not stated to be independent, so I will use the given joint probability in the general addition rule.” Or: “The process specifies independent outcomes, so the joint probability is the product of the individual probabilities; I will subtract that overlap to find the union.” A formula without the justification for its overlap may not support the conclusion.
Key Takeaway
When a question asks for an inclusive “or,” first determine the overlap. The general addition rule works for any two events; independence is one possible way to find the overlap, not a default assumption. Disjoint events have no overlap, but when both events have positive probability, they are not independent.
Check Your Understanding
For each question, identify the relevant event or overlap and explain whether independence is justified.
- Suppose \(P(A)=0.20\), \(P(B)=0.50\), and \(P(A\cap B)=0.10\). Find \(P(A\cup B)\). What would the product \(P(A)P(B)\) represent if the events were independent?
- A model gives \(P(C)=0.40\) and \(P(D)=0.25\), but gives no joint probability and says nothing about independence. Can you find \(P(C\cup D)\) exactly? Explain.
- Two events are disjoint and have positive probabilities. Explain why multiplying their probabilities does not give their actual overlap.
- A student calculates \(P(E\cup F)=P(E)P(F)\). Name the main error and describe the information needed to find the union correctly.
- A chance process explicitly states that \(G\) and \(H\) are independent, with \(P(G)=0.60\) and \(P(H)=0.20\). Find their overlap and then their inclusive union. Show why the product belongs in the overlap calculation.