Table 2. Neyman-optimized allocation as a function of sample size and stratum.

A
	nh[h] = sample per size per stratum[h]
sample size	nh[1]	nh[2]	nh[3]	nh[4]	∑(nh)
330	75	67	72	116	330
660	102	103	110	345	660
990	126	134	94	636	990
1320	143	138	128	911	1320
1650	114	1	1	1534	1650
B
	Nh[h] = total houses/stratum[h]
sample size	Nh[1]	Nh[2]	Nh[3]	Nh[4]	∑(Nh)
330	694	649	439	197	1979
660	569	611	454	345	1979
990	445	520	378	636	1979
1320	314	380	374	911	1979
1650	180	134	131	1534	1979
C
	πh = h[h]/Nh[h]
	sample size	h = 1	h = 2	h = 3	h = 4
	330	0.11	0.10	0.16	0.59
	660	0.18	0.17	0.24	1.00
	990	0.28	0.26	0.25	1.00
	1320	0.46	0.36	0.34	1.00
	1650	0.63	0.01	0.01	1.00
D
	upper boundary limits (persons per residence)
	sample size	h = 1	h = 2	h = 3	h = 4
	330	8.50	14.50	24.50	86.00
	660	7.50	12.50	19.50	86.00
	990	6.50	10.50	14.50	86.00
	1320	5.50	8.50	11.50	86.00
	1650	4.50	5.50	6.50	86.00

Table 2a: Optimal samples per stratum as a function of sample size. Table 2b: Optimal allocation of residential structures per stratum as a function of sample size. Table 2c: The inclusion probability π_h = h[h]/Nh[h] as a function of sample size. Table 2d: The upper strata boundaries as a function of sample size.

Table 2a lists the number of residential structures to be sampled in each stratum for optimal stratification of the variable “persons per residential structure.” Table 2b is the total number of residential structures per stratum, while Table 2c specifies the ratios of samples per stratum divided by the total number of residential structures per stratum. These ratios are not constant for each sample size because the optimization was constrained by Neyman allocation, rather than proportional allocation. Table 2d lists the upper boundary limits as a function of sample size.