Table 6.

Percent increase in RMSE compared to before deletion estimate, 1000 replicate samples (n=150)

Generated Model for Y and R			Hot deck estimators				Weighting estimators			Complete Case
	[]Y	[]R	wrhd(x)	wshd(x)	uhd(x)	uhd(xz)	wrr(x)	urr(x)	urr(xz)
1	XZ	XZ	66	56	49	53	55	53	39	108
2	XZ	X + Z	77	71	57	45	68	62	35	115
3	XZ	X	60	52	59	63	48	48	48	89
4	XZ	Z	55	45	37	29	45	43	22	42
5	XZ	ϕ	40	31	42	37	26	26	25	29
6	X + Z	XZ	83	78	58	49	75	71	38	139
7	X + Z	X + Z	116	109	74	40	108	97	31	159
8	X + Z	X	57	49	67	52	47	43	43	91
9	X + Z	Z	61	60	36	26	56	53	18	49
10	X + Z	ϕ	43	33	51	39	30	29	26	31
11	X	XZ	57	48	49	47	45	50	37	107
12	X	X + Z	50	42	44	48	39	51	35	82
13	X	X	53	41	47	55	39	40	44	132
14	X	Z	33	28	28	26	24	26	17	34
15	X	ϕ	34	27	29	38	21	22	23	26
16	Z	XZ	90	82	70	62	78	84	50	50
17	Z	X + Z	99	93	65	51	90	105	38	53
18	Z	X	74	59	89	65	55	56	50	47
19	Z	Z	80	73	52	39	68	70	29	74
20	Z	ϕ	50	40	68	47	35	35	32	35
21	ϕ	XZ	59	48	53	53	46	48	40	32
22	ϕ	X + Z	47	41	46	46	39	43	34	28
23	ϕ	X	63	49	53	61	46	46	50	30
24	ϕ	Z	37	33	32	34	28	29	22	29
25	ϕ	ϕ	43	32	38	46	29	29	30	29
Mean Percent			61	53	52	46	50	50	34	65

Lowest percent increase in RMSE among hot deck methods shown in italics.