How will you draw the OC curve for a single sampling plan with sample size 5, if the acceptance number is c = 2, assuming the lot size to be large ? Explain.
15 marks · directive: Explain
The core of a good answer
For a large lot the number of defectives in the sample is Binomial(5, p), so the OC curve is Pa(p) = (1 − p)⁵ + 5p(1 − p)⁴ + 10p²(1 − p)³; tabulating it gives Pa = 0·991 at p = 0·1, 0·942 at 0·2, 0·837 at 0·3, 0·683 at 0·4, 0·500 at 0·5, 0·317 at 0·6, 0·163 at 0·7 and 0·058 at 0·8, and plotting Pa against p gives a falling S-shaped curve with Pa = 0·95 at p ≈ 0·19 and Pa = 0·10 at p ≈ 0·75.
An outline written by upsc.help, not an official UPSC answer; every fact checked by a separate review. How we work
What the question asks
Explain what an OC curve is, justify the binomial model (large lot means sampling without replacement is effectively with replacement; the hypergeometric reduces to the binomial), derive Pa(p) for n = 5, c = 2, tabulate it for a grid of p, describe the plot and read off the AQL, indifference and LTPD points. Comment on how discriminating such a small plan is.
Answer outline
Introduction
The OC (operating characteristic) curve of an acceptance plan shows the probability Pa of accepting a lot as a function of the lot fraction defective p. Plan: draw n = 5 items, accept if the number of defectives d ≤ c = 2.
Step 1: Model
- Lot size N large relative to n, so d ~ Binomial(n = 5, p) (for small lots, sampling without replacement gives the hypergeometric).
- Pa(p) = P(d ≤ 2) = Σ_{d=0}^{2} C(5, d) p^d (1 − p)^(5−d)
- = (1 − p)⁵ + 5p(1 − p)⁴ + 10p²(1 − p)³
Step 2: Tabulate
- p = 0·00: Pa = 1·000
- p = 0·05: Pa = 0·999
- p = 0·10: Pa = 0·991
- p = 0·20: Pa = 0·942
- p = 0·30: Pa = 0·837
- p = 0·40: Pa = 0·683
- p = 0·50: Pa = 0·500
- p = 0·60: Pa = 0·317
- p = 0·70: Pa = 0·163
- p = 0·80: Pa = 0·058
- p = 1·00: Pa = 0·000
Step 3: Draw the curve
- Horizontal axis p from 0 to 1 (or percent defective); vertical axis Pa from 0 to 1.
- Plot the tabulated points and join them smoothly: a decreasing reversed-S curve from (0, 1) to (1, 0), symmetric about p = 0·5 because c = 2 is the middle of 0 to 5.
- Mark the key points: Pa = 0·95 at p ≈ 0·189 (AQL for α = 0·05), Pa = 0·50 at p = 0·5 (indifference quality), Pa = 0·10 at p ≈ 0·753 (LTPD for β = 0·10).
Step 4: Interpretation
- The ideal OC curve is a vertical step at the acceptable quality level; this curve is very flat, so the plan discriminates poorly: even lots that are 30% defective are accepted 84% of the time.
- Increasing n (with c/n fixed) makes the curve steeper; decreasing c shifts it left and makes the plan stricter.
- With the Poisson approximation (mean 5p) the values differ noticeably here (for example 0·920 at p = 0·2) because n is small and p large, so the exact binomial should be used.
Conclusion
The OC curve of the (5, 2) plan is Pa(p) = (1 − p)⁵ + 5p(1 − p)⁴ + 10p²(1 − p)³, plotted from the table above; it shows high acceptance for lots up to about 20% defective and only weak protection against bad lots. Values computed in Python.
Facts to remember
- OC curve: Pa(p) = P(d ≤ c | n, p).
- Large lot: d ~ Binomial(n, p); small lot: hypergeometric.
- AQL point Pa = 1 − α; LTPD point Pa = β; indifference point Pa = 0·5.
- Larger n gives a steeper, more discriminating OC curve.