A population is divided into three strata. The following data are taken from the population :
Stratum h | N_h | S_h² | C_h
1 | 200 | 25 | 4
2 | 300 | 36 | 9
3 | 500 | 49 | 16
Determine the sample size for each stratum for a total sample of size 100 under (i) proportional allocation and (ii) optimum allocation.
10 marks · directive: Determine
The core of a good answer
(i) Proportional allocation n_h = nN_h/N gives 20, 30 and 50. (ii) Optimum allocation n_h ∝ N_hS_h/√C_h, with weights 500, 600 and 875 (total 1975), gives 25.32, 30.38 and 44.30, i.e. 25, 31 and 44 after rounding to total 100. Ignoring costs, Neyman allocation would give 16, 29 and 55.
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
Proportional allocation uses only the stratum sizes. Because a per-unit cost C_h is given, the optimum allocation is the cost-optimal rule n_h ∝ N_hS_h/√C_h, which minimises V(ȳ_st) for a fixed total cost (or cost for fixed variance). State the formula, compute the weights and round so the total is 100.
Answer outline
Introduction
N = 200 + 300 + 500 = 1000; S_h = 5, 6, 7; √C_h = 2, 3, 4; n = 100.
Step 1: Proportional allocation
- n_h = n N_h/N: n₁ = 100 × 0.2 = 20, n₂ = 100 × 0.3 = 30, n₃ = 100 × 0.5 = 50.
Step 2: Optimum allocation formula
- Cost C = c₀ + ∑C_h n_h; minimising V(ȳ_st) = ∑W_h²S_h²/n_h − ∑W_hS_h²/N subject to fixed cost gives n_h = n (N_hS_h/√C_h)/∑(N_hS_h/√C_h).
Step 3: Weights
- N₁S₁/√C₁ = 200 × 5/2 = 500; N₂S₂/√C₂ = 300 × 6/3 = 600; N₃S₃/√C₃ = 500 × 7/4 = 875; total 1975.
Step 4: Sample sizes
- n₁ = 100 × 500/1975 = 25.32; n₂ = 100 × 600/1975 = 30.38; n₃ = 100 × 875/1975 = 44.30.
- Rounded so the total is 100: 25, 31, 44 (the largest fractional part, stratum 2, takes the extra unit).
Step 5: Neyman allocation (equal costs) for comparison
- n_h ∝ N_hS_h = 1000, 1800, 3500 (total 6300): 15.87, 28.57, 55.56, i.e. 16, 29, 55.
Conclusion
Proportional: (20, 30, 50). Optimum with the given costs: (25, 31, 44). The high cost in stratum 3 pulls units away from it, even though it is the largest and most variable stratum. Figures computed in Python.
Examples and data to cite
- A numerical minimisation in Python of V(ȳ_st) at fixed total cost ∑C_hn_h returns the same proportions 25.32 : 30.38 : 44.30.
Facts to remember
- Proportional allocation: n_h = nN_h/N.
- Cost-optimum allocation: n_h ∝ N_hS_h/√C_h.
- Neyman allocation (equal costs): n_h ∝ N_hS_h.
- V(ȳ_st) = ∑W_h²(1 − f_h)S_h²/n_h, W_h = N_h/N.