When does the order of outcomes matter in a probability count?
Quick answer
Order matters when swapping the selected items changes the result you care about; it does not when the swap leaves the result unchanged. First define what counts as one outcome, then count both the full sample space and the favorable outcomes using that same definition. For two items chosen from four, that means 12 ordered results or 6 unordered pairs. Counting alone gives a probability only when outcome likelihoods are also handled correctly.
The question, the details, the next step.
How do I decide whether to count ordered results or just selections? I keep getting different answers.
Start by asking: if I swap the selected items, has the result I care about changed? If yes, count the two arrangements separately. If no, treat them as one result. For example, first- and second-place finishes are different if the runners swap places, while a group chosen for a committee with equal roles stays the same when its members are listed in another order. This is the distinction between permutations (ordered lists) and combinations (selections where order does not matter). [S1][S2]
Before counting, write a plain-language definition of one distinct outcome. This prevents you from using ordered counts for the whole sample space but unordered counts for the event—or the reverse.
Could you show both ways with a small example?
Suppose four distinct tokens are A, B, C, and D. You draw two without replacement. Assume for this example that each possible ordered draw is equally likely.
If the result records the first draw and the second draw, there are 4 choices for the first and 3 remaining choices for the second: 4 × 3 = 12 ordered results. A few are AB, BA, AC, and CA. AB and BA are different ordered results.
If you care only which two tokens were drawn, the results are the six pairs AB, AC, AD, BC, BD, and CD. Now AB and BA describe the same pair. Each unordered pair has two orders, so the 12 ordered results correspond to 6 pairs. Choosing 2 of 4 distinct items gives 12 ordered outcomes and 6 unordered pairs. [S1]
How does that change the probability if I want a result containing A?
Keep the same outcome definition for favorable results and for the whole sample space.
With ordered draws, the six favorable results are AB, AC, AD, BA, CA, and DA: A can appear first or second, and the other token can be any of the three. That gives 6 favorable results out of 12. With unordered pairs, the favorable results are AB, AC, and AD: 3 out of 6. Both counts give 1/2 under the stated equal-likelihood assumption.
Do not divide 3 unordered favorable pairs by 12 ordered outcomes. Those numbers describe different kinds of results. Also, counting outcomes this way does not establish that they are equally likely in every experiment; that must be justified by the setup. [S1]
What should I write down before doing a count so I can catch a mismatch?
Use a short definition before applying a formula:
“An outcome is [what the result records]. Swapping [items or positions] [does/does not] change that outcome. Therefore I will count [ordered results / unordered selections]. The total number is [total], and the number meeting the event is [favorable]. I am treating the outcomes as [equally likely / not established as equally likely] because [reason or assumption].”
For the token example, the unordered version is: “An outcome is the pair of tokens drawn. Swapping their draw order does not change the pair. I count unordered selections: 6 total, 3 containing A. Under the equal-likelihood assumption, the probability is 3/6.” If you cannot explain why the outcome counts have the likelihoods your calculation assumes, pause: ordered-versus-unordered counting alone does not settle the probability.
A practical checklist
State exactly what information one outcome records.
Ask whether swapping the selected items changes that result.
Use the same ordered or unordered definition for total and favorable outcomes.
For a small example, list outcomes to check for duplicates or omissions.
Check whether equal likelihood is justified before using a favorable-count fraction.
A few more useful answers
Why do the ordered and unordered counts give the same probability here?
Every unordered pair corresponds to two ordered draws, so both the total and favorable counts are doubled in the ordered version. Under the example’s equal-likelihood assumption, the ratio stays the same. [S1]
Can I use the number of favorable outcomes divided by the total in every problem?
Only when the counted outcomes are equally likely, or when the calculation otherwise accounts for their different likelihoods. A count by itself does not establish equal likelihood. [S1]
What if swapping the items matters for one part of the question but not another?
Define the result for the particular event you are calculating. If the question asks only which items were selected, ignore order; if it asks their sequence or ranking, preserve order. [S2]
Sources
References used for this guide.
- [S1] OpenStax Contemporary Mathematics, Section 7.6: Probability with Permutations and Combinations openstax.org · Checked
- [S2] OpenStax Contemporary Mathematics, Section 7.3: Combinations openstax.org · Checked