Table 1.
Case Study 1 (Cross-Sectional): ψ = 1.5 (β3=0.4055) 2-sided test with size (α=0.05) and power (1−β=0.80) | |||||||
---|---|---|---|---|---|---|---|
Design | Random Sampling | Covariate Dependent Sampling (Dependent on Z) | Outcome Dependent Sampling | ||||
Cost | Sample Sizes | Cost | Sample Sizes | Cost | Sample Sizes | ||
Classic X | $15,970,150; n = 7210 | ||||||
2-Stage Z_Quest/X ρ=0.288 | $8,730,190 (54.7%) | n0 = 96646 n1= 869 |
$8,710,604 (54.5%) | n0= 92734 n1=990 |
$7,350,710 (46.0%) | n0= 11751 n1= 1335 |
|
2-Stage Z_SF/X ρ=0.892 | $5,518,340 (34.6%) | n0= 9505 n1= 90 |
$8,305,180 (52.0%) | n0=9390 n1=135 |
$7,082,560 (44.3%) | n0= 7595 n1= 380 |
|
Case Study 2 (Cross-Sectional): ψ = 1.5 (β3=0.4055) 2-sided test with size (α =0.05) and power (1−β =0.80) | |||||||
Design | Random Sampling | Covariate Dependent Sampling (Dependent on E) | Outcome Dependent Sampling | ||||
Cost | Sample Sizes | Cost | Sample Sizes | Cost | Sample Sizes | ||
Classic X | $13,446,000; n = 7210 | ||||||
2-Stage Z_Quest/X ρ=0.288 | $8,405,620 (62.5%) | n0= 93266 n1= 991 |
$5,641,020 (42.0%) | n0= 53382 n1=1029 |
$2,503,640 (18.6%) | n0 = 16992 n1 = 721 |
|
2-Stage Z_SF/X ρ=0.892 | $5,015,670 (37.3%) | n0 = 9489 n1= 96 |
$4,918,830 (36.6%) | n0 = 9218 n1 =128 |
$4,424,380 (32.9%) | n0= 7595 n1 = 380 |
|
Case Study 3 (Longitudinal): β3=−0.01 2-sided test with size (α=0.05) and power (1−β=0.80) | |||||||
Design | Random Sampling | Covariate Dependent Sampling (Dependent on Z) | Outcome Dependent Sampling | ||||
Cost | Sample Sizes | Cost | Sample Sizes | Cost | Sample Sizes | ||
Classic X | $25,189,225; n = 11,321 | ||||||
2-Stage Z_Quest/X ρ=0.288 | $15,477,600 (61.4%) | n0=100,000 n1=3,431 |
$15,491,200 (61.5%) | n0=99,998 n1 =3,438 |
$25,443,626 (101.1%) | n0 =12,254 n1 =11,202 |
|
2-Stage Z_SF/X ρ=0.892 | $8,503,830 (33.8%) | n0 =14,483 n1=107 |
$12,795,000 (50.8%) | n0=14,385 n1=155 |
$12,794,824 (50.8%) | n0=14,389 n1=152 |