Multivariable regression analysis of list experiment data on abortion: results from a large, randomly-selected population based study in Liberia

Table 2 Results for test of no design effect assumption. Table contains estimates of the population proportion reporting each number of items, and at least each number of items, by treatment group

	Number of list items reported
	0	1	2	3	4
Treatment list	0.040	0.278	0.446	0.168	0.068
Proportion at least	1.000	0.960	0.682	0.237	0.068
Control list	0.031	0.376	0.486	0.107	0.000
Proportion at least	1.000	0.969	0.593	0.107	0.000
Row 2 – Row 4	0.000	−0.009	0.089	0.130	0.068

A negative proportion in the bottom row suggests that the proportion reporting at least j items in the treatment group is less than the proportion reporting at least j items in the control group (Pr(Y > =j | T = 1) - Pr(Y > =j|T = 0) for j = 1,…, J). (J = 3, number of control item), and could be consistent with evidence for a design effect

ISSN: 1478-7954