Table 4. Neyman stratification of Bo sections by “residential structures per section” and “persons per section.”.

			Stratification variable: “residential structures per section”		Stratification variable: “persons per section”
(1) Section	(2) Residential structures per section	(3) Persons per section	(4) allocation = (1, 1, 1, 1)	(5) allocation = (2, 1, 6, 1)	(6) allocation = (1, 1, 1, 1)	(7) allocation = (2, 1, 6, 1)
Roma	4	139	1	1	1	1
Moibawo Farm	17	135	1	1	1	1
Dodo	26	597	1	1	1	2
Bo Central	33	273	1	1	1	1
Toubu	34	454	1	1	1	2
Kpetewoma	46	640	1	2	1	2
Komende	56	1103	2	2	2	3
Salina	59	580	2	2	1	2
Reservation	66	637	2	2	1	2
Kindia Town	102	1160	2	3	2	3
Lewabu	105	879	2	3	2	3
New York	116	1088	3	3	2	3
Njai Town	127	2298	3	3	3	4
New Site south	136	1248	3	3	2	4
Tengbewabu	136	1068	3	3	2	3
Yemoh Town	152	1858	3	3	3	4
Kissi Town	154	2490	3	3	3	4
Kulanda Town	197	3882	4	4	4	4
Nduvuibu	205	2552	4	4	3	4
New London	208	2873	4	4	4	4

4-level Neyman stratification boundaries for the cluster (section) list stratified by the number of residential structures per section (cols. 4–5), and the number of persons per section (cols. 6–7). In practice, the latter may be unknown. An allocation of (1, 1, 1, 1) sections per stratum is optimal for a sample size of 4 sections, which was used in our examples. The (2, 1, 6, 1) allocation, shown for comparison, is optimal for a 10-section sample. See Table 3.