Construct a single sampling plan for attributes, given the following data :
A.Q.L. = 0·05
Producer’s risk = 0·05
Consumer’s risk = 0·10
LTPD = 0·20
10 marks · directive: Construct
The core of a good answer
With p₁ = 0·05, α = 0·05, p₂ = 0·20, β = 0·10 the operating ratio p₂/p₁ = 4; from Poisson unity values (np at Pa = 0·95 and Pa = 0·10) the smallest c whose ratio np(0·10)/np(0·95) does not exceed 4 is c = 5 (ratio 3·549), and n = 9·275/0·20 = 46·4, rounded up to 47 (holding β), or n = 2·613/0·05 = 52·3, rounded down to 52 (holding α), any n from 47 to 52 meeting both risks; the plan n = 47, c = 5 gives Pa(0·05) = 0·967 and Pa(0·20) = 0·093 (Poisson). An exact binomial search shows n = 38, c = 4 also meets both risks.
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
The plan (n, c) must pass through two points of the OC curve: Pa(AQL) ≥ 1 − α and Pa(LTPD) ≤ β. A standard method is the Poisson approximation with a table of unity values: compute the operating ratio p₂/p₁, choose c, then n. A good answer states the two conditions, shows the table step, gives the plan, and checks it by computing both acceptance probabilities.
Answer outline
Introduction
Given: AQL p₁ = 0·05 with producer's risk α = 0·05, so Pa(0·05) should be at least 0·95; LTPD p₂ = 0·20 with consumer's risk β = 0·10, so Pa(0·20) should be at most 0·10. Plan: inspect n items, accept the lot if defectives d ≤ c.
Step 1: Conditions on the OC curve
- Pa(p) = P(d ≤ c) with d ~ Binomial(n, p), approximated by Poisson(np) for small p and large lots.
- Need P(d ≤ c | np₁) ≥ 0·95 and P(d ≤ c | np₂) ≤ 0·10.
- Operating ratio R = p₂/p₁ = 0·20/0·05 = 4.
Step 2: Choose c from unity values (Poisson)
- For each c find m₁ with P(Poisson(m₁) ≤ c) = 0·95 and m₂ with P(Poisson(m₂) ≤ c) = 0·10; ratio m₂/m₁:
- c = 3: m₁ = 1·366, m₂ = 6·681, ratio 4·890
- c = 4: m₁ = 1·970, m₂ = 7·994, ratio 4·057
- c = 5: m₁ = 2·613, m₂ = 9·275, ratio 3·549
- The ratio must not exceed R = 4 for both risks to be met, so c = 5.
Step 3: Choose n
- Hold β exactly: n = m₂/p₂ = 9·275/0·20 = 46·4, round up to n = 47.
- Hold α: n = m₁/p₁ = 2·613/0·05 = 52·3; Pa(p₁) ≥ 0·95 needs n ≤ 52·3, so round down to n = 52.
- Any n from 47 to 52 satisfies both conditions under the Poisson approximation; the smallest, n = 47, is the economical choice.
Step 4: Check the plan n = 47, c = 5
- Poisson: Pa(0·05) = P(Poisson(2·35) ≤ 5) = 0·967 ≥ 0·95; Pa(0·20) = P(Poisson(9·4) ≤ 5) = 0·093 ≤ 0·10.
- Binomial: Pa(0·05) = 0·971, Pa(0·20) = 0·070; both risks met.
- Exact binomial search for the smallest plan meeting both conditions gives n = 38, c = 4 (Pa(0·05) = 0·960, Pa(0·20) = 0·099); mention this as a refinement.
Conclusion
Recommended plan: take a random sample of 47 items from each lot, accept if 5 or fewer are defective, otherwise reject (or screen). The producer's risk is about 3% and the consumer's risk about 9%, both within the stated limits. All values recomputed in Python (scipy).
Examples and data to cite
- Tabled plans such as MIL STD 105D focus on either the AQL or the LTPD; a two-point plan like this one, solved from two points on the OC curve, controls both.
Facts to remember
- Single sampling plan (n, c): accept the lot if the number of defectives d ≤ c.
- Pa(p) = Σ_{d=0}^{c} C(n, d) p^d (1 − p)^(n−d) ≈ Σ e^(−np)(np)^d/d!.
- Producer's risk α = 1 − Pa(AQL); consumer's risk β = Pa(LTPD).
- Operating ratio p₂/p₁ fixes c; then n = m/p from the unity value m.
- The β-based n (rounded up) and the α-based n (rounded down) bound the range of n meeting both risks; choosing within it trades α against β.