Course contentsShow
← Mathematics
Lesson 2914 of 3,12157. Common DistributionsFree lesson

Continuity Correction

Adjusting discrete values by ±0.5 when approximating discrete distributions with the normal.

Continuity Correction

What you'll learn: How to adjust discrete probability values by ±0.5 when using the continuous normal distribution to approximate discrete distributions like the binomial.

The Bridge Between Discrete and Continuous

When you use the normal approximation to the binomial (or other discrete distributions), you're replacing a staircase with a smooth curve. The problem? A discrete random variable takes exact integer values (0, 1, 2, 3, ...), but the normal distribution is continuous and spreads probability smoothly across intervals.

The continuity correction fixes this mismatch by treating each discrete value as if it occupies an interval. For example, the discrete value X = 5 is treated as the interval [4.5, 5.5] on the continuous scale.

When and How to Apply It

Basic rule: Adjust the boundary of your discrete value by ±0.5 in the direction that captures the probability you want.

  • P(X = k) becomes P(k − 0.5 < Z < k + 0.5)
  • P(X ≤ k) becomes P(Z < k + 0.5)
  • P(X < k) becomes P(Z < k − 0.5)
  • P(X ≥ k) becomes P(Z > k − 0.5)
  • P(X > k) becomes P(Z > k + 0.5)

Analogy: Histogram Bars

Imagine a histogram where each integer has a bar centered at that value, extending 0.5 units left and right. When you approximate with the normal curve, you're asking "what area under the curve covers these bars?" The correction ensures you capture the full bar width.

Example Intuition

If you want P(X ≤ 10) for a binomial, you're including all outcomes up through 10. On the continuous scale, extend to 10.5 to capture the "bar" centered at 10.

Key Takeaway: Continuity correction adjusts discrete boundaries by ±0.5 to accurately approximate discrete probabilities with the continuous normal distribution.