What Is an Operating Characteristic (OC) Curve and How Do You Use It in Acceptance Sampling?

When you receive a batch of parts from a supplier, you can't always inspect every single item. Acceptance sampling lets you decide whether to accept or reject an entire lot based on a small sample. But how confident are you in that decision? The Operating Characteristic (OC) curve—rooted in ISO 2859-1 and the classic Dodge-Romig sampling theory—answers this question by showing the probability that a lot will be accepted for any given level of quality.

What It Is

An OC curve is a graphical tool used in quality control to evaluate the performance of a sampling plan. It plots the probability of acceptance (Pa) on the vertical axis against the lot defect proportion (p) on the horizontal axis.

The curve helps you answer two critical questions:

• What is the chance of accepting a "good" lot (Producer's risk, α)
• What is the chance of accepting a "bad" lot (Consumer's risk, β)

A well-designed OC curve rises steeply near the acceptable quality limit (AQL), meaning good lots are almost always accepted, while poor lots are almost always rejected.

How It Works / Formula

For a single sampling plan defined by sample size n and acceptance number c (accept the lot if the number of defectives in the sample is ≤ c), the probability of acceptance is calculated using the binomial distribution (when lot size is large relative to n) or the hypergeometric distribution (for small lots).

Using the binomial approximation, the formula is:

\[
Pa(p) = \sum_{x=0}^{c} \binom{n}{x} p^x (1-p)^{n-x}
\]

Where:

• n = sample size
• c = maximum allowable number of defectives
• p = true proportion defective in the lot
• x = number of defectives found in the sample

By computing Pa(p) for various values of p (e.g., 0.01, 0.02, 0.05, 0.10), you can plot the OC curve.

A Worked Illustrative Example

Example data (illustrative only):

Suppose you use a sampling plan with n = 50 and c = 2. You want to know the probability of accepting a lot that actually contains 4% defective items (p = 0.04).

Using the binomial formula:

\[
Pa(0.04) = \sum_{x=0}^{2} \binom{50}{x} (0.04)^x (0.96)^{50-x}
\]

Calculating each term:

• x = 0: (0.96)^50 ≈ 0.1299
• x = 1: 50 × 0.04 × (0.96)^49 ≈ 0.2706
• x = 2: 1225 × 0.0016 × (0.96)^48 ≈ 0.2762

Summing these:

\[
Pa(0.04) ≈ 0.1299 + 0.2706 + 0.2762 = 0.6767
\]

So, there is approximately a 67.7% chance of accepting a lot with 4% defectives under this plan. If you repeated this calculation for p = 0.01, 0.02, 0.05, and 0.10, you would trace out the full OC curve and see how the plan discriminates between acceptable and unacceptable quality.

Common Pitfalls

• Ignoring lot size: For small lots, the hypergeometric distribution is more accurate than the binomial. Using the wrong model can mislead you.
• Misreading producer vs. consumer risk: The curve near p = AQL reflects producer risk (rejecting good lots); the curve near the lot tolerance percent defective (LTPD) reflects consumer risk (accepting bad lots). Don't confuse the two.
• Assuming a perfect curve: No sampling plan is perfect. The OC curve always shows some overlap—there is no point where acceptance drops instantly from 100% to 0%.

---

Ready to visualize your own sampling plan? Use the free OC Curve calculator at https://www.6sq.com/tools/oc_curve/ to generate the curve for any n and c in seconds.