Why Counting Rules Matter
A finite experiment can have many possible outcomes even when each individual step is simple. For example, a hypothetical claim-submission process might offer several submission routes, and some routes might then require a choice of appointment time and report format. A systematic count helps describe the experiment’s sample space without listing every complete outcome.
Two basic rules organize these counts. Use addition when a complete outcome is chosen from one of several mutually exclusive alternatives. Use multiplication when a complete outcome is formed by making a sequence of choices. The key is to define what counts as one complete outcome before applying either rule.
Here \(|A|\) means the number of elements in a finite set \(A\). The addition rule requires disjointness: no outcome may be counted in two of the sets being added. The multiplication rule concerns complete sequences of choices, not a claim that the choices are statistically independent. It applies to counting whenever the stated number of available choices at each stage is correct.
The Addition Principle: Choose One Alternative
Suppose a complete outcome must belong to exactly one of several distinct categories. If category \(j\) contains \(n_j\) possible outcomes, count the outcomes in each category and add. The categories must not overlap, or an outcome that appears in more than one category would be counted more than once.
This is the counting version of organizing a sample space into disjoint parts. It is often helpful to write down the alternatives in words first: “route A or route B or route C.” The word “or” signals addition only after the alternatives have been checked to be mutually exclusive.
Worked Example: Counting Submission Plans by Route
A hypothetical insurer allows a claimant to submit a document package in one of three ways. The online route has 5 distinct package templates. The staffed-call route has 3 appointment windows and 2 language options. The mail route has 4 distinct packet types. Count the possible complete submission plans, where a plan includes the chosen route and every choice required within that route.
Method: split the plans into disjoint route categories. A plan cannot be both an online submission and a staffed-call submission under the stated process, so the three route categories are disjoint.
The online route contributes \(5\) plans.
For the staffed-call route, a plan includes both an appointment window and a language option. There are \(3\) choices for the window and \(2\) for the language, giving \(3\times2=6\) plans.
The mail route contributes \(4\) plans. Therefore, the total is $$ 5+6+4=15. $$ There are 15 complete submission plans.
Check: listing the route totals as \(5,6,4\) accounts for every route and counts each plan once. A common mistake would be to add \(3+2\) for the staffed-call route, which counts its stage choices rather than its complete window-language pairs.
The Multiplication Principle: Complete a Sequence
When a complete outcome is made by filling successive positions or stages, count the available choices at each stage. If every possible partial outcome has the same number of choices at the next stage, multiply the stage counts. The choices need not be independent in a probability sense; the condition is about how many options are available after each partial sequence.
For instance, choosing a desk, then a form, then a follow-up category creates a complete outcome only when all three choices have been made. Counting the first-stage options alone does not count complete outcomes.
Worked Example: Constructing a Review Record
For a hypothetical statistical audit, a record is formed by choosing one of 4 claim categories, one of 3 review durations, and one of 2 report formats. Every duration is available for every claim category, and both formats are available for every category-duration pair. How many distinct complete records can be formed?
Method: multiply the choices across successive stages. There are \(4\) choices for the claim category. For each category, there are \(3\) available durations, so there are \(4\times3=12\) category-duration pairs. For each such pair, either of \(2\) report formats can be chosen.
Thus the number of complete records is $$ 4\times3\times2=12\times2=24. $$ There are 24 distinct records.
Check: each record specifies exactly one category, duration, and format. The result is consistent with 12 category-duration pairs, each extended by 2 formats. The common mistake avoided here is adding \(4+3+2=9\): that counts the choices listed at the stages, not the combinations that complete all stages.
Why the Multiplication Rule Holds
The multiplication principle follows by grouping complete outcomes according to their first choice. Each first-stage choice determines a group of outcomes. If there are \(n_1\) first-stage choices and each group contains \(n_2\) outcomes from the second stage, then the groups are disjoint and together contain \(n_1n_2\) outcomes. Extending this grouping stage by stage gives the product for any finite number of stages.
The fixed-count condition matters. If different branches lead to different numbers of later options, there is no single stage count to multiply across all branches. Count each branch using multiplication, then add the disjoint branch totals. A table or tree diagram makes this adjustment transparent.
Tables and Trees for Unequal Branches
A tree diagram starts with the first choice and draws a new branch for each possible next choice. Continue until each path represents one complete outcome. The number of terminal paths is the desired count. A table can summarize the same information when the branches are easy to list.
For a branch with \(a\) choices at one stage and \(b\) choices after each of them, that branch contributes \(ab\) complete outcomes. If there are several mutually exclusive first-stage branches, add their contributions. This “multiply within branches, add across branches” method works even when the options differ from one branch to another.
Worked Example: Counting Records When Options Vary by Desk
A hypothetical review center assigns a record to one of three intake desks. Desk A offers two forms: form X has 3 follow-up categories, while form Y has 2. Desk B offers three forms, P, Q, and R, and each has 2 follow-up categories. Desk C offers one form, Z, with 4 follow-up categories. Count all complete desk-form-follow-up records.
Method: multiply within each desk branch, then add the disjoint desk totals. The desk choices are mutually exclusive, but the form and follow-up options differ by desk.
| Desk branch | Forms and follow-up choices | Complete records |
|---|---|---|
| A | X: 3; Y: 2 | \(3+2=5\) |
| B | P: 2; Q: 2; R: 2 | \(2+2+2=6\) |
| C | Z: 4 | \(4\) |
For Desk A, the branches through X and Y contribute \(1\times3=3\) and \(1\times2=2\), respectively, for \(5\) records. For Desk B, its three form branches contribute \(1\times2\) each, for \(6\) records. Desk C contributes \(1\times4=4\) records. The total is $$ 5+6+4=15. $$ There are 15 complete records.
Check: the desk totals are disjoint, and the form-specific counts have all been included. The mistake to avoid is treating every desk as though it offered the same number of follow-up choices and multiplying one average or one branch count across all desks.
Choosing the Correct Rule
A practical way to decide what to do is to describe a complete outcome in one sentence, then ask how it is built.
State exactly what choices or labels distinguish one outcome from another.
If an outcome belongs to one of several disjoint categories, addition applies across those categories. If it requires a choice at each successive stage, multiplication applies along a path.
If they do, multiply the stage counts. If they vary by earlier choices, count each branch separately and add the branch totals.
Verify that every complete outcome appears in one branch and only one branch.
A count of outcomes is not a probability. The rules above determine how many outcomes are in a finite sample space or in a specified group of outcomes. They do not show that the outcomes are equally likely. A probability calculation requires a probability model in addition to a count.
Common Exam Traps
- Adding choices from successive stages. If a record needs a category and a format, the complete record is a pair, not one choice from a combined list. Multiply when each pair is allowed.
- Multiplying overlapping alternatives. If a complete outcome can belong to more than one category, first make the categories disjoint or account for overlap. The simple addition rule requires disjoint sets.
- Assuming every branch has the same size. When later options depend on earlier choices, inspect each branch rather than using one product for the entire tree.
- Counting partial outcomes as complete ones. A choice of desk alone is not a desk-form-follow-up record. Count only terminal paths that have all required stages.
- Confusing counting with probability. A larger number of outcomes in a category does not by itself establish its probability unless the probability model justifies an appropriate connection.
Check Your Understanding
For each problem, count complete outcomes, not just choices at an individual stage.
- A hypothetical intake system offers 6 online templates or 3 mail packet types, with exactly one route and one type selected. How many submission plans are there? (Answer: 9)
- A review file requires one of 5 regions, one of 2 schedules, and one of 3 display formats. Every choice is available after every earlier choice. How many files can be formed? (Answer: 30)
- A service center has two desks. Desk A offers 2 forms, each with 4 follow-up choices. Desk B offers 3 forms, each with 2 follow-up choices. How many complete desk-form-follow-up records are possible? (Answer: 14)
- A hypothetical report is submitted by either the web route or the phone route. The web route has 3 appointment options and then 2 formats per appointment. The phone route has 4 appointment options and then 3 formats per appointment. How many complete reports are possible? (Answer: 18)
- A finite experiment first selects one of 3 categories. In category 1 there are 2 second-stage choices, in category 2 there are 5, and in category 3 there are 4. Each second-stage choice leads to exactly 2 final choices. How many complete paths are there? (Answer: 22)