Normal Approximation: Binomial Guide, Formulas & Continuity Correction

What Is Normal Approximation?

The binomial distribution gives the exact probability of getting exactly k successes in n independent trials when the probability of success on each trial is p. The formula involves factorials that grow rapidly with n, making hand calculation impractical for large samples. The normal approximation sidesteps this by treating the binomial as if it were continuous, converting the target value to a Z-score and using the standard normal distribution table.

The theoretical justification comes from the Central Limit Theorem: the sum of many independent, identically distributed random variables converges to a normal distribution as the number of terms grows. A binomial random variable is the sum of n independent Bernoulli trials, so it follows that the binomial converges to normal as n increases. This is why the approximation improves with larger samples.

Conditions for Normal Approximation

The approximation is only trustworthy when the underlying distribution is reasonably symmetric. For a binomial distribution, that happens when both tails have enough expected events to prevent severe skew. The standard check, used in most introductory statistics courses and on the AP Statistics exam, is called the Rule of 5.

Why the Conditions Matter

When p is very small or very large, the binomial distribution is skewed: most probability mass sits near zero or near n. In those cases, a symmetric bell curve is a poor fit. For example, with n = 20 and p = 0.05, np = 1 — well below 5. The distribution is strongly right-skewed. Forcing a normal approximation there produces meaningless probability estimates.

As p approaches 0.5, the binomial becomes symmetric, and the approximation improves even at smaller n. The Rule of 5 catches the worst cases; the Rule of 10 adds a margin for applied work where getting the right probability matters more than a quick calculation. Full coverage of this skew behavior is in the binomial distribution guide.

Normal Approximation Formulas

Three quantities are needed: the mean, the standard deviation, and the Z-score after applying the continuity correction.

Once you have the Z-score, look it up in the Z-table or use a calculator's normal cumulative distribution function. The table gives the area to the left of the Z-score. For probabilities in the right tail, subtract the table value from 1. For the area between two Z-scores, subtract the smaller table value from the larger one.

Continuity Correction

The binomial counts integers: 0, 1, 2, 3, and so on. The normal distribution covers every real number along a continuous line. When you map the integer k onto the continuous curve, you are treating a single point as if it had width. The continuity correction expands each integer to an interval of width 1 centered on it — specifically from k − 0.5 to k + 0.5 — which is the range of real numbers that would round to k.

Without the correction, you are reading the probability at the single point k, which has measure zero on a continuous distribution. With it, you read the area of a thin strip, which matches the discrete bar in the histogram. For large n, the difference is small. For moderate n near the condition threshold, it is noticeable enough to matter.

How to Apply Normal Approximation (8 Steps)

Worked Examples

Each example follows the 8-step procedure above. Arithmetic is shown in full so you can follow each number. All Z-score lookups use the standard four-decimal-place table distributed by NIST.

Example 1 — Airline No-Show Problem (P(X < 15))

Example 2 — Exact Probability: P(X = 45) from 100 Coin Flips

Example 3 — Poisson to Normal Approximation

Normal Approximation Calculator

Enter the binomial parameters, select the inequality type, and click Calculate. The calculator shows every intermediate step: conditions check, mean, standard deviation, continuity-corrected boundary, Z-score, and the final probability.

Method Comparison

Choosing between exact binomial computation and the normal approximation depends on sample size, the accuracy you need, and the tools available. The Poisson distribution is a third option when n is large and p is very small.

For a deeper comparison of the normal and t-distributions, see the T-distribution vs. normal distribution guide. Coverage of the empirical rule — a related application of normal distribution properties — is in the empirical rule article.

Common Mistakes to Avoid

Python and R Implementation

Software handles the calculation with far less risk of arithmetic error. Both examples below compute the approximated probability and the exact binomial result side by side so you can verify the approximation's accuracy.

Python (SciPy)

import numpy as np
import scipy.stats as stats

n, p, k = 100, 0.4, 35

# Normal approximation parameters
mu = n * p
sigma = np.sqrt(n * p * (1 - p))

# Continuity correction for P(X <= k)
k_corrected = k + 0.5
prob_approx = stats.norm.cdf(k_corrected, mu, sigma)

# Exact binomial for comparison
prob_exact = stats.binom.cdf(k, n, p)

print(f"mu = {mu}, sigma = {sigma:.4f}")
print(f"Approximated P(X ≤ {k}) = {prob_approx:.4f}")
print(f"Exact P(X ≤ {k})        = {prob_exact:.4f}")

n <- 100
p <- 0.4
k <- 35

mu <- n * p
sigma <- sqrt(n * p * (1 - p))

# Continuity correction for P(X <= k)
k_corrected <- k + 0.5
prob_approx <- pnorm(k_corrected, mean = mu, sd = sigma)

# Exact binomial for comparison
prob_exact <- pbinom(k, size = n, prob = p)

cat("mu =", mu, " sigma =", round(sigma, 4), "\n")
cat("Approximated:", round(prob_approx, 4), "\n")
cat("Exact:", round(prob_exact, 4), "\n")
  

For more on working with distributions in R, Khan Academy's normal approximation to binomial article provides a solid complementary reference. Penn State's STAT 414 course materials on approximating the binomial distribution cover the theoretical derivation in more depth.

Related Concepts on Statistics Fundamentals

FAQs