Use the normal approximation to the binomial distribution. Calculate probabilities, z-scores, and check validity conditions with continuity correction.
The Normal Approximation Calculator estimates binomial probabilities with the normal distribution when the trial count is large enough for the approximation to be reasonable.
It checks the usual validity conditions, applies continuity correction when requested, and turns a discrete count problem into a z-score calculation. That makes it useful whenever you want a quick range probability without summing many binomial terms by hand.
The page also shows the mean, standard deviation, and validity status so you can tell whether the approximation is appropriate before you rely on the result.
Binomial probabilities become tedious once the number of trials gets large, especially when you need a range probability rather than a single count. The normal approximation gives a fast estimate and makes the continuity correction explicit.
Showing the validity check alongside the probability helps you see whether the approximation is a good fit or whether you should switch to the exact binomial calculation.
Normal Approximation: Z = (X ± 0.5 - μ) / σ, where μ = np, σ = √(npq), q = 1 - p. The continuity correction ±0.5 adjusts for the discrete-to-continuous transition. Validity requires np ≥ 5 and nq ≥ 5.
Result: 0.7287 (approximately 72.87%)
For 100 coin flips with p = 0.5, μ = 50 and σ = 5. With continuity correction, P(45 ≤ X ≤ 55) ≈ P(44.5 ≤ X ≤ 55.5) using the normal curve, yielding about 0.7287.
The normal approximation to the binomial distribution is one of the most important applications of the Central Limit Theorem in statistics. When you have a binomial random variable X ~ Bin(n, p), and if np and nq are both sufficiently large, then X is approximately normally distributed with mean μ = np and standard deviation σ = √(npq).
This approximation transforms the tedious task of computing exact binomial probabilities — which involves factorials and powers — into a simple z-score lookup. The key insight is that as n grows, the shape of the binomial distribution increasingly resembles the iconic bell curve of the normal distribution.
Since the binomial distribution is discrete (only integer values) while the normal distribution is continuous, a correction factor of 0.5 is applied to improve accuracy. For P(X ≤ k), we compute P(Z ≤ (k + 0.5 - μ)/σ). For P(X ≥ k), we compute P(Z ≥ (k - 0.5 - μ)/σ). This half-unit adjustment bridges the gap between discrete and continuous probability models.
Without continuity correction, the approximation systematically underestimates or overestimates probabilities, particularly for smaller sample sizes. The correction becomes less important as n increases because the relative size of 0.5 compared to σ shrinks.
Normal approximation is widely used in hypothesis testing for proportions, confidence interval construction, quality control (monitoring defect rates), political polling (predicting election outcomes), clinical trials (evaluating drug effectiveness), and A/B testing in marketing. Understanding when the approximation is valid — and when it breaks down — is a fundamental skill in applied statistics.
Last updated:
The rule of thumb is that both np and nq must be at least 5 (preferably 10 or more). The calculator automatically checks these conditions and displays a validity indicator.
Continuity correction adds or subtracts 0.5 from the x value to account for approximating a discrete distribution with a continuous one. It generally improves accuracy and is recommended unless your instructor says otherwise.
Accuracy depends on n and p. For large n with p near 0.5, the approximation is excellent. For extreme p values (near 0 or 1), you may need a much larger n. The Poisson approximation may be better when p is very small.
The z-score tells you how many standard deviations a value is from the mean. A z-score of 0 means the value equals the mean. Values beyond ±2 are considered unusual, and beyond ±3 are rare.
Yes! For sample proportions, use n = sample size and p = population proportion. The approximation to the sampling distribution of p-hat uses the same underlying mathematics.
Both are validity thresholds. np ≥ 5 (and nq ≥ 5) is the minimum for acceptable approximation. np ≥ 10 (and nq ≥ 10) provides a better approximation. The calculator shows both quality levels.