Intro statistics · Discrete probability
Binomial distribution practice problems with answers
Binomial questions can go wrong in three places: using the binomial when the setup isn’t binomial, turning words like “at least” into the wrong inequality, and being off by one on the calculator. This guide covers those one at a time. It starts with a quick test for “is this binomial?”, takes the formula apart piece by piece, then works four full examples before a 14-question practice set you can check as you go.
Is it binomial? The BINS test
A binomial random variable X counts the number of successes in a set of repeated trials. You can only use it when all four of these are true. If even one fails, stop: the binomial formula will give you a wrong answer that looks perfectly reasonable.
Binary
Each trial has exactly two outcomes, which we call success and failure. “Success” is just the outcome you are counting. It can be something bad, like a defective part.
Independent
The result of one trial doesn’t change the chances on any other trial.
Number of trials is fixed
You know n before you start: 10 questions, 20 patients, 15 free throws.
Same probability
The chance of success, p, is the same on every trial.
When all four hold, we write X ~ B(n, p): X is binomial with n trials and success probability p.
Red flags: setups that look binomial but aren’t
| Setup | Which condition fails | Why |
|---|---|---|
| Pick 5 people from a group of 20 (no repeats) and count how many are left-handed | Independent and Same p | Sampling without replacement from a small group. Each pick changes who is left, so p shifts from draw to draw. (This is hypergeometric.) |
| Roll a die until you get a 6 and count the rolls | Fixed n | You don’t know the number of trials ahead of time. (This is geometric, see below.) |
| Roll a die 10 times and record the number shown each time | Binary | Six outcomes, not two. It becomes binomial only if you count one kind of result, like “number of 6s”. |
| A basketball player takes 20 shots and gets more accurate as she warms up | Same p (and maybe Independent) | The chance of success changes during the trials. |
The 10% rule. Real surveys almost always sample without replacement, so strictly speaking the trials are not independent. But if the sample is less than 10% of the population, the change in p is so small that the binomial is still a very good model. Polling 1,000 adults from a city of 400,000 is fine. Picking 5 people from a class of 20 (25% of the class) is not.
The formula, piece by piece
P(X = k) = C(n, k) · pk · (1 − p)n − k
It looks dense, but each piece answers one simple question.
| Piece | Plain English | Example: n = 3 coin flips, k = 2 heads, p = 0.5 |
|---|---|---|
| C(n, k) = n! ⁄ (k!(n − k)!) | How many different orders the k successes can show up in. Also written nCk or “n choose k”. | HHT, HTH, THH: 3 orders. C(3, 2) = 3! ⁄ (2! · 1!) = 3. |
| pk | The chance that the k chosen trials are all successes. | 0.52 = 0.25 |
| (1 − p)n − k | The chance that the other n − k trials are all failures. | 0.51 = 0.5 |
| Multiply | One order has probability pk(1 − p)n − k, and there are C(n, k) orders. | 3 × 0.25 × 0.5 = 0.375 |
Check it by listing: of the 8 equally likely outcomes (HHH, HHT, HTH, HTT, THH, THT, TTH, TTT), exactly 3 have two heads, and 3 ⁄ 8 = 0.375. The formula is just a fast way of doing that count when n is too big to list.
Two quick checks catch most formula errors: the two exponents must add up to n, and the answer must be between 0 and 1.
Exactly, at most, at least: the word translator
Most binomial mistakes happen before any math: the wording gets turned into the wrong inequality. Because X can only be a whole number, “fewer than 3” and “at most 2” are the same event. Use this table with n trials and a cutoff k.
| The question says | Inequality | How to get it | Example with k = 4 |
|---|---|---|---|
| exactly k | X = k | One formula term (pdf) | P(X = 4) |
| at most k, no more than k | X ≤ k | cdf at k | P(X ≤ 4) |
| fewer than k, less than k | X ≤ k − 1 | cdf at k − 1 | P(X ≤ 3) |
| at least k, no fewer than k | X ≥ k | 1 − cdf at k − 1 | 1 − P(X ≤ 3) |
| more than k, greater than k | X ≥ k + 1 | 1 − cdf at k | 1 − P(X ≤ 4) |
| between a and b, inclusive | a ≤ X ≤ b | cdf at b − cdf at a − 1 | P(2 ≤ X ≤ 4) = P(X ≤ 4) − P(X ≤ 1) |
Swipe the table sideways to see every column.
A good habit: write the possible values of X as a number line from 0 to n and circle the ones the question wants. If n = 10 and the question says “more than 4”, you circle 5, 6, 7, 8, 9, 10. The ones you didn’t circle, 0 to 4, are the complement, which is exactly what the calculator gives you.
The “at least one” shortcut. “At least one” means everything except zero, so
P(X ≥ 1) = 1 − P(X = 0) = 1 − (1 − p)n
No combinations, no adding terms. You only need the chance that every single trial fails.
Mean and standard deviation
Mean (expected value)
μ = np
Standard deviation
σ = √(np(1 − p))
The mean is the long-run average count: if you repeated the whole set of n trials many times, the number of successes would average out to np. It doesn’t have to be a whole number. With n = 10 and p = 0.25, μ = 2.5 even though you can never get 2.5 successes in one round.
The standard deviation tells you how much the count typically varies from one round to the next. A count more than about 2 standard deviations from the mean is unusual. Don’t forget the square root: np(1 − p) by itself is the variance.
TI-84, Excel and Google Sheets
Your calculator or spreadsheet does the adding for you. There are only two functions to know, and the whole trick is that the cumulative one always means “at most”.
| You want | TI-84 (2nd, VARS for DISTR) | Excel and Google Sheets |
|---|---|---|
| P(X = k) | binompdf(n, p, k) | =BINOM.DIST(k, n, p, FALSE) |
| P(X ≤ k) | binomcdf(n, p, k) | =BINOM.DIST(k, n, p, TRUE) |
| P(X ≥ k) | 1 − binomcdf(n, p, k − 1) | =1 − BINOM.DIST(k − 1, n, p, TRUE) |
| P(X > k) | 1 − binomcdf(n, p, k) | =1 − BINOM.DIST(k, n, p, TRUE) |
Watch the order of the inputs. The TI-84 takes n, p, k. Excel and Sheets take k, n, p, then the cumulative flag: FALSE for “exactly”, TRUE for “at most”. Typing the numbers in the wrong order is one of the most common ways to get a wrong answer from a correct plan.
Shape and the normal approximation
The shape of a binomial distribution depends on p:
- p below 0.5: skewed right (most of the bars sit at small counts, with a tail to the right).
- p = 0.5: exactly symmetric.
- p above 0.5: skewed left.
As n grows, the skew fades and the bars start to look like a bell curve. Try it: change n and p below and watch the shape, the mean and the standard deviation.
Binomial shape explorer
Mean np = 3. Standard deviation √(np(1 − p)) = 1.45. Shape: skewed right, because p is below 0.5 (the skew gets milder as n grows). Normal approximation: np = 3 and n(1 − p) = 7, so the rule of thumb (both at least 10) is not met.
The normal approximation, briefly
When n is large, adding up dozens of binomial terms by hand is painful, so many courses approximate the binomial with a normal curve that has mean np and standard deviation √(np(1 − p)). A common rule of thumb is that this works well when
np ≥ 10 and n(1 − p) ≥ 10
(some textbooks use 5 instead of 10). Because the binomial jumps in whole numbers and the normal curve is smooth, you move the cutoff by 0.5. This is the continuity correction: P(X ≥ 30) becomes P(Y ≥ 29.5), and P(X ≤ 30) becomes P(Y ≤ 30.5). Example 4 shows how much it helps. With software, you can always get the exact binomial answer instead.
Binomial vs geometric
Both models use independent yes-or-no trials with the same p. The difference is what you count.
| Feature | Binomial | Geometric |
|---|---|---|
| X counts | Successes in n trials | Trials until the first success |
| Fixed n? | Yes | No, you stop at the first success |
| Possible values | 0, 1, 2, …, n | 1, 2, 3, … (no upper limit) |
| Formula | C(n, k) pk(1 − p)n − k | (1 − p)k − 1 p |
| Example | Number of 6s in 10 rolls | Number of rolls until the first 6 |
Quick geometric example: the chance the first 6 comes on the third roll is (5⁄6)2(1⁄6) ≈ 0.116, two misses then a hit. A question with “until” or “first” in it is a strong hint that it’s geometric, not binomial.
Four worked examples
Example 1: Guessing on a quiz (exactly)
A quiz has 10 multiple-choice questions with 4 options each. You guess on every one. What is the probability you get exactly 3 right?
- Check BINS. Right or wrong (binary). Guesses don’t affect each other (independent). n = 10 is fixed. p = 1⁄4 = 0.25 on every question. Binomial: X ~ B(10, 0.25).
- Translate. “Exactly 3” means P(X = 3).
- Pieces. C(10, 3) = 10! ⁄ (3! · 7!) = 120. 0.253 = 0.015625. 0.757 ≈ 0.133484.
- Multiply. 120 × 0.015625 × 0.133484 ≈ 0.2503.
- Check. Exponents 3 + 7 = 10 = n. On a TI-84,
binompdf(10, 0.25, 3)= 0.2503.
Answer: about 0.250, a 1 in 4 chance. If you forget C(10, 3), you get 0.0021, which is the chance of one specific order, like getting exactly questions 1, 2 and 3 right.
Example 2: Defective chargers (at least one)
8% of phone chargers from a factory are defective. A store opens a box of 12 chargers from a huge shipment. What is the probability at least one is defective?
- Check BINS. Defective or not. The 12 are a tiny fraction of a huge shipment (well under 10%), so treat them as independent with the same p. X ~ B(12, 0.08), where “success” means defective.
- Translate. “At least one” is P(X ≥ 1). Adding P(1) through P(12) would take 12 terms. Use the complement.
- Complement. P(X = 0) = 0.9212 ≈ 0.3677.
- Subtract. P(X ≥ 1) = 1 − 0.3677 = 0.6323.
Answer: about 0.632. Even with only 8% defective, a box of 12 has a defective charger more often than not. A common wrong answer is 12 × 0.08 = 0.96. That is the expected number of defectives, not a probability.
Example 3: Free throws (at least, and fewer than)
A player makes 70% of free throws. In a game she shoots 15, and we assume the shots are independent. (a) What is P(she makes at least 12)? (b) What is P(she makes fewer than 10)?
- Model. X ~ B(15, 0.7).
- (a) Translate. At least 12 means X = 12, 13, 14 or 15. The complement is X ≤ 11, so P(X ≥ 12) = 1 − P(X ≤ 11).
- Calculator.
1 − binomcdf(15, 0.7, 11)≈ 1 − 0.7031 = 0.2969. - Check by adding the four terms. P(12) ≈ 0.1700, P(13) ≈ 0.0916, P(14) ≈ 0.0305, P(15) ≈ 0.0047. Sum ≈ 0.2968, which matches up to rounding.
- (b) Translate. Fewer than 10 means X ≤ 9. Calculator:
binomcdf(15, 0.7, 9)≈ 0.2784.
Answers: (a) about 0.297, (b) about 0.278. The off-by-one trap: 1 − binomcdf(15, 0.7, 12) gives 0.1268, which is P(X ≥ 13). It leaves out 12.
Example 4: Left-handed students (mean, SD, and the normal approximation)
Suppose 12% of students at a university of 30,000 are left-handed. In a random sample of 200 students, how many lefties do you expect, and how unusual would 30 or more be?
- Model. 200 is less than 10% of 30,000, so X is approximately B(200, 0.12).
- Mean. μ = np = 200 × 0.12 = 24 left-handed students.
- Standard deviation. σ = √(200 × 0.12 × 0.88) = √21.12 ≈ 4.60.
- Interpret. 30 is (30 − 24) ⁄ 4.60 ≈ 1.3 standard deviations above the mean. That is on the high side, but not unusual.
- Binomial model. P(X ≥ 30) =
1 − binomcdf(200, 0.12, 29)≈ 0.1176. - Normal check. np = 24 and n(1 − p) = 176 are both at least 10, so the approximation is allowed. With the continuity correction, z = (29.5 − 24) ⁄ 4.596 ≈ 1.20, and the normal curve gives P(Y ≥ 29.5) ≈ 0.116, close to the exact 0.118. Without the correction, z = (30 − 24) ⁄ 4.596 ≈ 1.31 and the normal curve gives about 0.096, which is noticeably worse.
Answer: expect about 24, give or take about 4.6. There is roughly a 12% chance of 30 or more, so it would not be surprising.
3-day free trial, no card needed.
Exam traps
| Trap | What it looks like | The fix |
|---|---|---|
| Off by one with the cdf | P(X ≥ 12) typed as 1 − binomcdf(15, 0.7, 12) | The cdf includes its cutoff. For “at least k”, subtract the cdf at k − 1. |
| Percent instead of decimal | binompdf(10, 25, 3) for a 25% chance | p must be between 0 and 1: type 0.25. An “answer” above 1 or an error message is the giveaway. |
| Forgetting the combination | 0.253 × 0.757 = 0.0021 | That’s one order. Multiply by C(n, k) to count all the orders. |
| Mixing up the input order | =BINOM.DIST(10, 3, 0.25, FALSE) | Excel and Sheets go k, n, p. The TI-84 goes n, p, k. |
| Mean instead of probability | “P(at least one) = np = 0.96” | np is a count. Probabilities never go above 1. Use 1 − (1 − p)n. |
| Forcing a binomial | Choosing 5 of 20 classmates and using p = 12⁄20 for every pick | Run BINS first. Without replacement from a small group, p changes. |
Practice: 14 binomial problems
Work each one on paper first, then check it. You get instant feedback and can try as many times as you like. Give probabilities as decimals to 3 or 4 places (answers within 0.002 count). Every question has a full worked solution underneath. Press Enter in a number box to check it.
Round 1: Is it binomial?
1. A class has 20 students, 12 of them women. You choose 4 students at random, without repeats, and let X be the number of women chosen. Is X binomial with n = 4 and p = 0.6?
Show solution
C. On the first pick P(woman) = 12⁄20 = 0.60. If the first pick is a woman, the second pick has 11⁄19 ≈ 0.58. Because p changes and the picks depend on each other, X is not binomial. The 10% rule doesn’t rescue it, since 4 is 20% of 20.
2. You roll a fair die until you get a 6, and X is the number of rolls it takes. Which model fits X?
Show solution
B. The rolls are independent with the same p = 1⁄6, but X counts trials until the first success, with no fixed n. That is the geometric distribution. For example, P(X = 3) = (5⁄6)2(1⁄6) ≈ 0.116.
3. A pollster randomly samples 1,000 adults from a city of 400,000, where 36% of adults own a dog. X is the number in the sample who own a dog. Is X (approximately) binomial?
Show solution
A. B: dog or no dog. I: a random sample that is 1,000 ⁄ 400,000 = 0.25% of the city, well under 10%, so the picks are close enough to independent. N: 1,000. S: p is approximately 0.36 for each person. So X is approximately B(1000, 0.36).
Round 2: Translate the words
4. A question asks for the probability of “fewer than 3” successes. Which event is that?
Show solution
B. Fewer than 3 is 0, 1 or 2 successes, so X ≤ 2. On a TI-84 that is binomcdf(n, p, 2).
5. X ~ B(12, 0.3). Which expression equals P(more than 4 successes)?
Show solution
D. “More than 4” means 5, 6, …, 12. The values not circled are 0 to 4, so P(X > 4) = 1 − P(X ≤ 4) = 1 − binomcdf(12, 0.3, 4) ≈ 1 − 0.7237 = 0.2763.
Round 3: Compute the probability
6. Exactly. Each seed in a packet sprouts with probability 0.4, independently. You plant 8 seeds. Find P(exactly 3 sprout).
Show solution
- X ~ B(8, 0.4), and we want P(X = 3).
- C(8, 3) = 8! ⁄ (3! · 5!) = 56.
- 0.43 = 0.064 and 0.65 = 0.07776.
- 56 × 0.064 × 0.07776 ≈ 0.2787. TI-84:
binompdf(8, 0.4, 3).
7. At most. 15% of a store’s customers use a coupon. Of the next 12 customers (independent), find P(at most 2 use a coupon).
Show solution
- X ~ B(12, 0.15). At most 2 means X ≤ 2.
- P(0) = 0.8512 ≈ 0.1422.
- P(1) = 12 × 0.15 × 0.8511 ≈ 0.3012.
- P(2) = 66 × 0.152 × 0.8510 ≈ 0.2924.
- Total ≈ 0.7358. TI-84:
binomcdf(12, 0.15, 2).
8. At least. A treatment works for 60% of patients. In a trial with 10 independent patients, find P(at least 8 respond).
Show solution
- X ~ B(10, 0.6). At least 8 means X = 8, 9 or 10.
- P(8) = 45 × 0.68 × 0.42 ≈ 0.1209.
- P(9) = 10 × 0.69 × 0.4 ≈ 0.0403.
- P(10) = 0.610 ≈ 0.0060.
- Total ≈ 0.1673. TI-84:
1 − binomcdf(10, 0.6, 7).
9. At least one. Each scratch-off ticket wins a prize with probability 0.10, independently. You buy 6. Find P(at least one wins).
Show solution
- X ~ B(6, 0.1). We want P(X ≥ 1).
- Complement: P(X = 0) = 0.96 = 0.531441.
- P(X ≥ 1) = 1 − 0.531441 ≈ 0.4686.
10. More than. A student guesses on all 20 questions of a test with 4 options each. Find P(more than 7 correct).
Show solution
- X ~ B(20, 0.25). More than 7 means X ≥ 8.
- The complement is X ≤ 7:
binomcdf(20, 0.25, 7)≈ 0.8982. - P(X > 7) = 1 − 0.8982 ≈ 0.1018. Excel:
=1-BINOM.DIST(7,20,0.25,TRUE).
Round 4: Mean, standard deviation and tools
11. 36% of adults in a large city own a dog. In a random sample of 50, what is the mean number of dog owners?
Show solution
μ = np = 50 × 0.36 = 18 dog owners. In plenty of samples of 50 you would see more or fewer, but they average out to 18.
12. Same setting: n = 50, p = 0.36. What is the standard deviation of the number of dog owners? Round to 2 or 3 decimals.
Show solution
- Variance: np(1 − p) = 50 × 0.36 × 0.64 = 11.52.
- σ = √11.52 ≈ 3.39.
- So a sample of 50 typically has about 18 ± 3.4 dog owners, and fewer than about 11 or more than about 25 (2 SDs away) would be unusual.
13. Which calculator command. X ~ B(15, 0.3). Which TI-84 command gives P(at least 4)?
Show solution
B. At least 4 is 4, 5, …, 15. The leftover values are 0 to 3, so P(X ≥ 4) = 1 − P(X ≤ 3) = 1 − binomcdf(15, 0.3, 3) ≈ 1 − 0.2969 = 0.7031.
14. Which spreadsheet formula. In Excel or Google Sheets, which formula gives P(X ≤ 5) for X ~ B(20, 0.3)?
Show solution
C. BINOM.DIST takes (number of successes, number of trials, probability, cumulative). TRUE means “at most”, so =BINOM.DIST(5, 20, 0.3, TRUE) ≈ 0.4164.
Score: 0 of 14 correct.
FAQ
How do I know if a problem is binomial?
Check four things, often remembered as BINS: each trial has two outcomes (Binary), the trials are Independent, the Number of trials is fixed in advance, and the Success probability is the Same on every trial. If any one fails, the count is not binomial. Sampling without replacement can be modeled approximately by a binomial if the sample is under 10% of the population.
What is the formula for binomial probability?
P(X = k) = C(n, k) × p^k × (1 − p)^(n − k). C(n, k) counts how many orders the k successes can appear in, p^k is the chance of the k successes, and (1 − p)^(n − k) is the chance of the n − k failures.
How do you find 'at least' with binomcdf?
binomcdf(n, p, k) gives P(X ≤ k), so use the complement. P(X ≥ k) = 1 − binomcdf(n, p, k − 1). For example, P(X ≥ 4) with n = 15 and p = 0.3 is 1 − binomcdf(15, 0.3, 3).
What is the shortcut for the probability of at least one success?
Use the complement: P(at least one) = 1 − P(none) = 1 − (1 − p)^n. With n = 12 and p = 0.08, that is 1 − 0.92^12 ≈ 0.632.
What are the mean and standard deviation of a binomial distribution?
The mean is np and the standard deviation is the square root of np(1 − p). With n = 200 and p = 0.12, the mean is 24 successes and the standard deviation is about 4.60.
When can you use the normal approximation to the binomial?
A common rule of thumb is np ≥ 10 and n(1 − p) ≥ 10 (some books use 5). When it applies, use a normal curve with mean np and standard deviation √(np(1 − p)), and add or subtract 0.5 at the cutoff (the continuity correction).
What is the difference between binomial and geometric distributions?
A binomial counts the number of successes in a fixed number of trials. A geometric counts how many trials it takes to get the first success, so there is no fixed n. Both need independent trials with the same p.
Keep practicing
Try 3 days free3-day free trial, no card needed.
Related guides
Sources and fact checks
Definitions, formulas and calculator steps were checked against these references. Every number on this page, including all answers in the practice set, was recomputed with a short script. The scenarios are made up for practice.
- OpenStax Introductory Statistics 2e, Section 4.3: Binomial Distribution
- OpenStax Introductory Statistics 2e, Section 4.4: Geometric Distribution
- OpenStax Introductory Statistics 2e, Section 4.5: Hypergeometric Distribution
- Statistics LibreTexts: Binomial Distribution (OpenIntro)
- Microsoft Support: BINOM.DIST function
- Google Docs Editors Help: BINOMDIST (BINOM.DIST in Google Sheets)