Binomial probabilities include a combination because exactly k successes can happen in several different positions. The power terms give the probability of one fixed pattern. The combination counts all matching patterns. For three fair coin tosses, exactly two heads can appear as HHT, HTH, or THH, so the probability is 3/8.
The factor C(n, k), read as “n choose k,” often looks like an extra step. It’s actually the step that changes the question. “Heads, heads, tails” names one sequence. “Exactly two heads” accepts every sequence with that count. Learning to separate those questions makes the formula easier to understand.
This lesson builds the counting factor from small examples you can check by hand. You’ll see why a combination belongs in a problem with ordered trials, how the pieces fit together, and when the model stops fitting. For other counting and probability tools, visit the statistics and probability hub.
Why Isn’t One Pattern Enough?
One pattern answers a question about a specific order, while exactly two heads allows three orders. With three independent fair tosses, each listed pattern has probability (1/2)(1/2)(1/2) = 1/8. Adding HHT, HTH, and THH gives 3/8. Using only 1/8 leaves out two outcomes that meet the question.
Imagine recording each toss in its own box. HHT means the first two boxes contain heads and the last contains tails. HTH moves the tail to the middle. THH puts the tail first. The records differ, even though every record has the same number of heads.
These three outcomes cannot happen together in one set of tosses. Your record can match only one row, so their probabilities add without overlap. Since all three probabilities equal 1/8, adding them is the same as multiplying 1/8 by three. That is the counting factor in action.
The word “exactly” also rules out HHH. It has at least two heads, but it has three heads rather than two. Before doing any calculation, write a sample outcome that qualifies and another that doesn’t. This quick check often reveals whether the problem asks for one pattern, one count, or several counts.
What Does The Combination Count?
The combination counts choices of success positions among the trial positions. For five trials with exactly two successes, choose two positions from 1, 2, 3, 4, and 5. There are ten choices. Once those positions are chosen, every remaining position must be a failure, so each choice defines one complete pattern.
You can list the position pairs: {1, 2}, {1, 3}, {1, 4}, {1, 5}, {2, 3}, {2, 4}, {2, 5}, {3, 4}, {3, 5}, and {4, 5}. For example, {2, 5} produces F S F F S. Here S means success and F means failure.
Why doesn’t {5, 2} create another pattern? Both pairs select the same boxes. Writing the chosen positions in a different order doesn’t change which trials succeed. The trials themselves still have an order. A combination works because we’re selecting a set of positions within that ordered record.
The usual count is C(n, k) = n!/[k!(n − k)!]. For five positions and two successes, this becomes 5!/(2!3!) = 10. If you select the two positions one at a time, you first count 5 × 4 = 20 choices. Dividing by 2 removes the repeated listings of each pair.
You can check the position count with the permutation and combination calculator. Enter the number of trials as n and the desired successes as the number selected. Use its combination result. That result counts patterns; it doesn’t yet give their probability.
Building The Binomial Formula
The binomial formula multiplies one pattern’s probability by the number of matching patterns. Let n be the trial count, k the success count, and p the success probability on each trial. Then P(X = k) = C(n, k)pk(1 − p)n − k, as presented by OpenStax’s binomial distribution lesson.
The first power contains one factor of p for each success. The second contains one factor of 1 − p for each failure. Independence lets us multiply these trial probabilities. Every pattern with the same count uses the same factors, although the factors appear in different orders.
For example, S F S F has probability p(1 − p)p(1 − p). F S F S has probability (1 − p)p(1 − p)p. Multiplication gives the same result for both: p2(1 − p)2. Changing success positions changes the record, but it doesn’t change that pattern’s probability.
Equal probability matters within the group you’re adding. You don’t need every possible result to be equally likely. If success has probability 0.2, four successes are less likely than four failures. Yet all patterns containing exactly one success still share a probability, which makes the combination shortcut work.
How Does The Formula Work With Four Trials?
With four independent trials and success probability 0.2 each time, exactly one success has probability 0.4096. A particular pattern, such as S F F F, has probability (0.2)(0.8)3 = 0.1024. The success can occupy any of four positions, so multiplying by C(4, 1) = 4 gives the total.
The accepted patterns are SFFF, FSFF, FFSF, and FFFS. Each includes one factor of 0.2 and three factors of 0.8. No two patterns describe the same complete result. Their sum is 0.1024 + 0.1024 + 0.1024 + 0.1024 = 0.4096, or 40.96%.
Exactly zero successes also has probability 0.4096 in this example. Its only pattern is FFFF, with probability (0.8)4. This equality happens because four possible one-success patterns balance their smaller individual probabilities against the single all-failure pattern. Different trial counts or success probabilities won’t generally give the same equality.
At least one success asks a broader question. It accepts one, two, three, or four successes. The only excluded pattern is FFFF, so use its complement: 1 − (0.8)4 = 0.5904. That’s 59.04%, not 40.96%. Reading the event correctly matters as much as entering the numbers correctly.
When Can You Use This Counting Shortcut?
Use the ordinary binomial formula when the trial count is fixed, the trials are mutually independent, each trial has two defined outcome categories, and success has the same probability each time. These conditions follow OpenStax’s binomial model. Together, they make every pattern with a given success count equally probable.
“Success” is just the outcome you’re counting. It doesn’t have to mean something good. You could define success as rolling a one and failure as rolling anything else. A die has six faces, but those faces fit into two categories for that question. Define your categories before choosing p.
Drawing objects without replacement usually breaks independence and the constant-probability condition. Suppose a bag holds two red and two blue counters. A red first draw changes the next red probability from 2/4 to 1/3. You must use the changing conditional probabilities rather than repeating 1/2 for every draw.
Independent trials can also have different success probabilities. If the first trial uses 0.2 and the second uses 0.6, then SF has probability (0.2)(0.4) = 0.08. FS has probability (0.8)(0.6) = 0.48. There are still two patterns with one success, but multiplying either probability by two gives the wrong total.
Boundary Cases And Probability Checks
Zero successes and all successes each have just one pattern, so C(n, 0) = C(n, n) = 1. The first record contains only failures; the second contains only successes. The factorial formula agrees because 0! = 1. See why zero factorial equals one for the empty-selection explanation.
If there are zero trials, zero successes is the only possible count. There is one empty record, with probability 1. If p = 0, every trial fails. If p = 1, every trial succeeds. Handle those cases directly; you don’t need a general claim that 00 is valid everywhere.
A negative success count, a count larger than n, or a noninteger count describes an impossible event here. Its probability is zero. Three tosses can’t produce four heads or one and a half heads. Identify that impossibility before trying to put invalid values into the factorial expression.
Every final probability must lie from zero to one. Also, the probabilities for all possible success counts must add to one, because every complete outcome has exactly one count. For three fair tosses, the following table groups all eight possible records without losing or repeating any record.
| Heads | Pattern count | Probability |
|---|---|---|
| 0 | 1 | 1/8 |
| 1 | 3 | 3/8 |
| 2 | 3 | 3/8 |
| 3 | 1 | 1/8 |
The probabilities add to 8/8 = 1. The pattern counts add to eight. These are separate checks: one checks the total probability, and the other checks the outcome count. Passing both gives you a useful way to catch a missing pattern or an extra counting factor in small examples.
Common Questions About Binomial Probabilities
Does Order Matter In A Binomial Experiment?
Yes, trial order distinguishes complete records such as HHT and HTH. The event “exactly two heads” accepts both records. A combination chooses the success positions without counting a repeated listing of those positions. The guide to permutations, combinations, and order explains the wider counting distinction.
Should I Multiply By A Combination For A Specified Sequence?
No. If the question specifies SFFF, calculate that pattern’s probability alone. In the four-trial example, it is 0.1024. Multiplying by four would answer a different question: exactly one success anywhere. Underline any words that fix positions, such as “first,” “last,” or “in this order,” before selecting a method.
Can A Probability Calculator Replace The Counting Step?
A calculator can check arithmetic, but first decide what event you’re counting. The probability calculator divides favorable outcomes by total equally likely outcomes. You can use it to check 3/8 for fair tosses. For unequal outcome probabilities, keep the binomial pattern count and each pattern’s probability visible.
The combination factor answers one question: how many success-position patterns match the requested count? The powers answer another: how likely is each such pattern? Write one accepted sequence, check the model conditions, and count its allowed position choices. Then multiply those two pieces to find exactly k successes.