Why Unordered Selections Need a Different Count
When a group is selected from a collection of distinct objects, the key question is whether the order of selection matters. If a committee has three members, listing its members as Ana, Bo, and Chen describes the same committee as listing Chen, Ana, and Bo. Counting those lists separately would count one committee several times. A combination counts the group once, regardless of the order in which its members could be listed.
This distinction is useful in finite counting problems throughout statistics. For example, a hypothetical actuary might select policies for a review sample. If the question asks which policies are in the sample, their order of selection is irrelevant. If it asks which policy is reviewed first, second, and so on, order matters. The objects and sample size may be identical, but the counting method changes with the question.
From Ordered Selections to Combinations
The earlier material on factorials and ordered selections gives the number of ways to choose and arrange \(r\) distinct objects from \(n\) as \(n!/(n-r)!\). This counts ordered lists: swapping two selected objects produces a different list. To count unordered groups instead, remove the overcounting caused by all the possible orders within each group.
Each group of \(r\) distinct objects has exactly \(r!\) different orderings. For instance, a group of three objects can be listed in \(3!=6\) orders. Every ordered selection belongs to exactly one unordered group, and each such group accounts for \(r!\) ordered selections. Dividing the ordered count by \(r!\) therefore counts each group once.
Here \(0!=1\), so the formula also gives \(\binom{n}{0}=1\): there is exactly one way to select nothing. Similarly, \(\binom{n}{n}=1\), because there is exactly one way to select all \(n\) objects. These boundary cases are useful checks on calculations.
A Step-by-Step Counting Method
Before calculating, describe what counts as one outcome. If an outcome is just a set or group, use combinations. If it includes positions, a ranking, or assigned roles, some or all of the choices may need to be counted as ordered. When a selection has separate requirements for distinct categories of objects, count the choices within each category and multiply them, provided each complete selection is formed by one choice from each category.
Confirm that the objects are distinct and determine how many are selected.
Ask whether swapping two selected objects changes the outcome described in the problem.
For an unordered selection from one collection, use \(\binom{n}{r}\). For separate group requirements, use a product of combination counts.
Verify that each eligible group is counted once and that no group violates a stated requirement.
Worked Example: Selecting a Mixed Review Committee
A hypothetical insurer forms a five-person review committee from five pharmacists and seven actuaries. The committee must contain two pharmacists and three actuaries. How many different committees can be formed? Committee members have no separately assigned roles.
Method. The outcome is an unordered committee. Choose two of the five pharmacists, then choose three of the seven actuaries. Each permissible pair of choices determines exactly one committee.
The pharmacist selections number \( \binom{5}{2} = 5!/(2!3!) = (5 \times 4)/(2 \times 1) = 10 \). The actuary selections number \( \binom{7}{3} = 7!/(3!4!) = (7 \times 6 \times 5)/(3 \times 2 \times 1) = 35 \).
By the multiplication principle, the total is \(10 \times 35 = 350\) committees.
Check. Each committee contains exactly two pharmacists and three actuaries, and changing the order in which its five members are listed does not create a new committee. The common mistake avoided here is treating the five committee members as an ordered list.
Selections Followed by an Order or Assignment
A problem can contain both an unordered selection and an ordered task. First count the possible groups; then count the ways each group can be arranged or assigned, if the problem distinguishes those outcomes. The order should not be included until the question actually makes it relevant.
Worked Example: Choosing an Audit Sample and a Lead
A hypothetical audit team has nine distinct contracts available for review. It selects four contracts as a sample, then designates one of those four as the lead contract for the review schedule. How many sample-and-lead outcomes are possible?
Method. First count the unordered samples, then count the possible lead choices within each sample. The sample count is \( \binom{9}{4} = 9!/(4!5!) = (9 \times 8 \times 7 \times 6)/(4 \times 3 \times 2 \times 1) = 3024/24 = 126 \).
For every sample of four contracts, there are four possible choices for the lead. Therefore the number of sample-and-lead outcomes is \(126 \times 4 = 504\).
Check by an alternative count. The lead can be selected first in 9 ways. The other three sample contracts can then be selected from the remaining 8 in \( \binom{8}{3} = 8!/(3!5!) = (8 \times 7 \times 6)/(3 \times 2 \times 1) = 56 \) ways. This gives \(9 \times 56 = 504\), agreeing with the first method.
If the problem instead asked for the four contracts in a fully ordered sequence, the count would be \(9 \times 8 \times 7 \times 6 = 3024\). That is not the same question: choosing a sample and a lead does not specify an order for the other three contracts. The common mistake avoided is counting every possible ordering of the sample when only one member has a designated role.
Applying Requirements to Particular Members
A fixed-member requirement can often be handled by holding that member in the selection and choosing the remaining places from the other objects. This changes the number of available objects and the number of places still to fill. Check both quantities carefully: a required member occupies one place, so that place should not be counted again.
Worked Example: A Required Volunteer in a Training Group
A training group of five volunteers is selected from twelve people: four are trained and eight are new. The group must contain exactly two trained volunteers and must include a particular trained volunteer, Mira. How many groups satisfy both requirements?
Method. Mira is already one of the two required trained volunteers. There are three other trained volunteers available, and one more trained volunteer must be chosen, giving \( \binom{3}{1} = 3!/(1!2!) = 3 \) choices.
The group needs five people in total. With two trained members, it must include three new volunteers. There are eight new volunteers, so the number of ways to choose them is \( \binom{8}{3} = 8!/(3!5!) = (8 \times 7 \times 6)/(3 \times 2 \times 1) = 56 \).
The choices are made from separate categories, so the total number of groups is \(3 \times 56 = 168\).
Check. Every counted group includes Mira, one other trained volunteer, and three new volunteers. It therefore has five distinct members and exactly two trained members. The common mistake avoided is choosing two additional trained volunteers after including Mira, which would produce three trained members rather than the required two.
Conditions, Symmetry, and Common Traps
The combination formula is a count of subsets of distinct objects. It is not a universal formula for every situation involving “choosing.” If an object may be selected more than once, the assumption that the \(r\) selected objects are distinct no longer describes the problem, so the formula must not be applied automatically. Likewise, if the roles or positions of selected objects are distinguishable, a plain combination count may omit outcomes.
A helpful symmetry check is \( \binom{n}{r}=\binom{n}{n-r} \). Choosing which \(r\) objects to include determines the same group as choosing which \(n-r\) objects to leave out. This identity can make a factorial calculation shorter, but it does not change the meaning of the counted groups.
When reading a problem, use its description of the outcome rather than the verb alone. “Select four contracts” usually describes a group, but “select four contracts and assign them to four distinct review days” describes an assignment as well. In a multi-part count, state what each factor represents; this makes it easier to notice a missing requirement or an unnecessary ordering.
Combinations form a basic counting tool for finite models and are aligned with the SOA VEE Mathematical Statistics syllabus. Their value lies not just in memorizing the formula, but in deciding whether two lists represent the same outcome. That decision determines whether order should be counted or removed.
Check Your Understanding
For each problem, identify the selection being counted and check whether order matters before calculating.
- A quality team chooses three distinct sensors from a batch of eight for inspection. How many unordered groups of sensors can it choose? (Answer: \( \binom{8}{3}=56 \))
- A hypothetical product review panel needs two engineers from a group of six and two actuaries from a group of five. No roles are assigned within the panel. How many panels can be formed? (Answer: \( \binom{6}{2}\binom{5}{2}=15 \times 10=150 \))
- A four-person committee is chosen from ten distinct employees and must include a particular employee. How many committees are possible? (Answer: \( \binom{9}{3}=84 \))
- A display uses three distinct tokens chosen from seven, and the chosen tokens are then placed in three ordered positions. How many different displays can be made? (Answer: \( \binom{7}{3}3!=35 \times 6=210 \))
- A five-person group is chosen from four trainees and seven mentors. The group must contain exactly two trainees. How many groups meet the requirement? (Answer: \( \binom{4}{2}\binom{7}{3}=6 \times 35=210 \))