A question describes several attempts, each with success or failure. That sounds binomial, but two outcomes alone are not enough. You also need a fixed number of trials, independence and a constant probability of success under the model.
Once those conditions fit, the calculation follows a clear structure. Define what is counted, identify the number of trials and success probability, and translate the requested event before using a calculator.
This guide uses original numerical examples to explain that process. Match practice questions to your current syllabus and component. Students reviewing Probability and Statistics 1 can explore AS Level Math S1 at Abeel Eschool for available lesson coverage.
Define a success without judging the outcome
In probability, a success is the event being counted. It could be a defective item, a missed target or a delayed delivery. The word does not mean that the outcome is desirable.
Suppose X is the number of defective items among five independently inspected items. If each item has defect probability 0.4 under the assumed model, a success is “defective,” n = 5 and p = 0.4.
Write that definition before the distribution notation. Saying only “X is defective” does not make clear whether X is a count, a category or the result for one item.
A useful definition identifies the quantity, the trial group and the relevant outcome. It prevents a later switch from counting defective items to counting acceptable items without changing p.
Check the four modelling conditions
First, the number of trials is fixed in advance. Second, each trial is classified into two outcomes for the event being studied. Third, trials are independent. Fourth, the probability of success remains constant.
These conditions describe the probability model, not merely the appearance of the question. Five attempts do not automatically form five independent trials if the outcome of one changes the conditions of the next.
For instance, drawing several counters without replacement from a small bag changes the remaining composition. The success probability can change after each draw, so a simple binomial model may not fit.
Likewise, trials that continue until the first success do not have a fixed trial count in the same way as a binomial count. Read the stopping rule and assumptions before selecting a distribution.
Write the parameters and possible values
For the illustrative inspection model, write X ~ B(5, 0.4). X can take integer values from zero through five. It cannot be negative, exceed five or take a fractional count such as 2.5.
The second parameter is p, not the expected number of successes. Here the expected count is 5 × 0.4 = 2, which is different from the probability 0.4.
Check that p lies between zero and one. If it is given as a percentage, convert it correctly: forty per cent is 0.4, not 40 or 0.04.
Naming the parameters is a quick way to expose uncertainty before a long calculation. If you cannot state n and p confidently, return to the description of a trial and its success event.
Understand the exact-probability expression
For X ~ B(n,p), the probability of exactly r successes is nCr × p^r × (1−p)^(n−r). Each part has a meaning.
The factor p^r accounts for r successes. The factor (1−p)^(n−r) accounts for the remaining failures. The combination nCr counts the different positions in which those r successes can occur.
For X ~ B(5,0.4), exactly two successes has probability 5C2 × 0.4² × 0.6³. This is 10 × 0.16 × 0.216 = 0.3456.
Calculating only 0.4² × 0.6³ gives the probability of one particular arrangement of successes and failures. The combination factor includes every qualifying arrangement, assuming the binomial conditions hold.
Translate inequality language before using a function
For an integer count, “at least two” means X ≥ 2. “More than two” means X ≥ 3. “At most two” means X ≤ 2, while “fewer than two” means X ≤ 1.
Write the inequality beside the phrase. This small step prevents an otherwise correct calculator method from including or excluding the wrong endpoint.
The event X ≤ 2 includes zero, one and two successes. The event X = 2 includes only two. A cumulative-probability function and an exact-probability function therefore answer different questions.
Be particularly careful when subtracting from one. The complement of X ≥ 2 is X ≤ 1, not X ≤ 2. Mark the allowed count values on a short number line if the boundary is unclear.
Use a complement for “at least one”
With X ~ B(5,0.4), the probability of at least one success is one minus the probability of no successes. The no-success probability is 0.6⁵ = 0.07776.
The required result is therefore 1 − 0.07776 = 0.92224. You could add the probabilities of one, two, three, four and five successes, but the complement is shorter here.
The complement works because “at least one” and “none” are mutually exclusive and together cover every possible count. Do not apply the same shortcut to another event without identifying its actual complement.
For example, “at least two” excludes both zero and one. Its probability is one minus the sum of those two probabilities, rather than one minus the probability of zero alone.
Build a cumulative answer from exact probabilities
For the same model, P(X = 0) is 0.07776. P(X = 1) is 5C1 × 0.4 × 0.6⁴ = 0.2592.
Therefore P(X < 2) = P(X ≤ 1) = 0.07776 + 0.2592 = 0.33696. The probability of at least two successes is 1 − 0.33696 = 0.66304.
This short calculation gives a useful manual check for a calculator result. If your cumulative function returns a different value, inspect the endpoint, parameters and whether the function means “up to and including.”
When several terms are involved, keep sufficient precision and round only as required at the end. Repeatedly rounding small terms can introduce avoidable differences.
Interpret expectation and variance separately
For a binomial random variable, the mean is np and the variance is np(1−p). In the example, the mean is 2 and the variance is 1.2.
The mean describes an expected count over repeated instances of the model. It does not mean that every group of five contains exactly two successes. A particular group can contain any allowed count.
Variance measures spread, and standard deviation is the square root of variance. If the question asks for standard deviation, reporting 1.2 would confuse two different quantities.
The expected value need not be an integer. A count variable can have an expectation such as 2.4 even though no individual outcome can equal 2.4 successes. Average behaviour and possible individual values are different ideas.
Read calculator inputs as part of the mathematics
Before evaluating a distribution function, identify where the calculator expects n, p and the target count. Know whether the selected function returns P(X = r) or a cumulative probability.
Test the function on a small example you can calculate by hand. A familiar manual case can reveal whether an entry convention or endpoint has been misunderstood.
Record the event and a clear expression in your written solution rather than submitting only a decimal. This makes the method traceable if a numerical entry goes wrong.
Use the facilities allowed for your actual examination and practise on your own permitted calculator. Do not assume that another student’s button sequence applies to every model.
Spot a model that does not fit
Suppose five items are selected without replacement from a box containing only a small fixed number of defective and acceptable items. Once an item is removed, both the remaining total and possibly the defective count change.
That dependence needs to be considered. A sequence of conditional probabilities or another suitable counting method may be appropriate instead of applying a binomial expression automatically.
Another warning sign is a success probability that deliberately changes by trial. A task may become harder after each success, or the question may state different conditions on different attempts.
Do not override an explicit modelling assumption without reason, but do not silently add independence where it is absent. A good solution says why the selected model fits the facts supplied.
Use checks that challenge the final answer
A probability must lie between zero and one. A cumulative probability P(X ≤ r) cannot decrease when r increases. The probabilities of all possible exact counts should sum to one, allowing for rounding.
Check boundary cases. P(X ≤ n) is one and P(X > n) is zero. An impossible count should not produce a positive probability in the stated model.
Also check whether you counted successes or failures. Changing the counted event changes p and may change the interpretation of the requested inequality.
These checks cannot prove that every answer is correct, but they often reveal wrong endpoints or swapped parameters before the mistake reaches the final line.
Make revision target the earliest mistake
Record whether an incorrect attempt began with model choice, parameter selection, event translation, the combination factor or numerical evaluation. Each cause needs a different follow-up.
If you chose the wrong distribution, practise classifying short scenarios without calculating. If the model was right but the inequality was wrong, translate phrases into count lists. If the calculator entry failed, compare an exact and cumulative calculation for the same small model.
Use Abeel’s Cambridge past-paper guide for independent attempts and reattempts. The AS/A2 planning guide can help fit focused revision around other topics.
Explore Abeel Eschool and current courses for relevant learning options. Ask about suitable support through admissions and FAQs, sharing your attempted event definition and parameters.
Frequently asked questions
Are two possible outcomes enough for a binomial model?
No. The usual model also needs a fixed number of trials, independence and a constant success probability.
Does success have to mean something positive?
No. It means the event being counted, such as a defect or a missed attempt. Define it before choosing p.
Why is there a combination factor?
It counts the different positions of the specified number of successes. One product of success and failure probabilities describes only one arrangement.
Is “more than two” the same as “at least two”?
No. For an integer count, more than two begins at three, while at least two includes two.
Can the binomial mean be non-integer?
Yes. The mean is an expected count across repeated situations. Individual outcomes remain integer counts.
Ask about O and A Level courses
Share your subjects, current study stage and the support you need. The Abeel Eschool team can help you check the current course options.