When is it appropriate to use a Normal distribution to approximate a Binomial distribution?

It is appropriate when $n$ is large and $p$ is close to $0.5$. Specifically, the conditions $np > 5$ and $n(1-p) > 5$ must both be satisfied to ensure the distribution is sufficiently bell-shaped and symmetrical.

What are the parameters $\mu$ and $\sigma^2$ for the Normal distribution used in the approximation?

The mean is $\mu = np$ and the variance is $\sigma^2 = np(1-p)$. These are derived directly from the expected value and variance formulas of the original Binomial distribution.

Why is a continuity correction necessary when approximating a Binomial distribution?

Because the Binomial distribution is discrete (values are points) while the Normal distribution is continuous (values are areas). The correction of $\pm 0.5$ allows the continuous model to account for the 'width' of the discrete integer bars.

How does $P(X < 10)$ in a Binomial distribution change when applying a continuity correction for a Normal approximation?

It becomes $P(X_N < 9.5)$. Since $X < 10$ does not include 10, the largest integer included is 9; the continuity correction must therefore cover the bar for 9, which ends at 9.5.

How does $P(X \le 10)$ differ from $P(X < 10)$ in terms of continuity correction?

$P(X \le 10)$ becomes $P(X_N < 10.5)$ because it includes the bar for 10. $P(X < 10)$ becomes $P(X_N < 9.5)$ because it excludes the bar for 10 and only goes up to the end of the bar for 9.

What is a common error regarding the variance in the Z-score formula for this topic?

Students often use the variance $np(1-p)$ directly in the denominator of the Z-score formula instead of the standard deviation. The correct formula is $Z = \frac{x - \mu}{\sqrt{np(1-p)}}$.

What happens to the approximation if $p$ is very small (e.g., 0.01) even if $n$ is large?

The distribution becomes heavily skewed to the right, failing the $np > 5$ condition. In such cases, the Normal approximation will be inaccurate because the Normal distribution is perfectly symmetrical and cannot model high skewness.

If a question asks for $P(X = 15)$ using a Normal approximation, what range is calculated?

You calculate $P(14.5 < X_N < 15.5)$. This captures the entire area of the discrete bar centered at 15, which is necessary because the Normal probability of a single point is zero.

What is the 'inclusion rule' for continuity correction at the lower bound, such as $P(X \ge k)$?

To include $k$, you must start the continuous range at the beginning of the bar for $k$, which is $k - 0.5$. Therefore, $P(X \ge k) \approx P(X_N > k - 0.5)$.

When calculating $P(a \le X \le b)$, what is the corrected Normal range?

The corrected range is $P(a - 0.5 < X_N < b + 0.5)$. This ensures that the discrete bars for both the lower bound $a$ and the upper bound $b$ are fully included in the area calculation.

Library Podcasts

Courses

Referral & Rewards

Revision Notes

AS-Level

Cambridge International Examinations

Maths

Probability And Statistics 1

Statistical Distributions

Normal Approximation of a Binomial Distribution

Summary

The Normal distribution can be used to approximate the Binomial distribution when the number of trials is large and the probability of success is near 0.5. This transition from a discrete to a continuous model requires specific parameter adjustments and the application of a continuity correction to maintain accuracy.

1. Definition & Core Concepts

A diagram showing a Binomial histogram (blue bars) being overlaid by a smooth Normal distribution curve (red line), illustrating how the continuous curve approximates the discrete bars.

2. Underlying Principles & Conditions

Conditions for Validity: The approximation is considered reliable only if the sample size $n$ is large and the probability $p$ is not too close to 0 or 1. Specifically, both $np > 5$ and $nq > 5$ (where $q = 1 - p$ ) must be satisfied.
Symmetry Requirement: If $p$ is close to 0.5, the Binomial distribution is naturally symmetrical, making the Normal curve (which is perfectly symmetrical) a better fit.
Law of Large Numbers: As $n$ increases, the discrete jumps between Binomial probabilities become relatively smaller, causing the distribution to converge toward the Normal shape.

3. Methods & Techniques: Calculating Parameters

4. The Continuity Correction

5. Key Distinctions

6. Exam Strategy & Tips