. 2019 Mar 14;61(4):1003–1019. doi: 10.1002/bimj.201800222

Table 4.

Count data simulation results, with Poisson dependent data generated from models (14) (boxes 1–3) and Poisson covariate data generated from (15)

	β₀				β₁				β₂
	Mean	mSE	eSE	Cov	Mean	mSE	eSE	Cov	Mean	mSE	eSE	Cov
Poisson – $β = (3, 0.1, 0.1)$
True value	3.00			0.95	0.10			0.95	0.10			0.95
Full data	3.00	0.01	0.01	0.94	0.10	0.01	0.01	0.95	0.10	0.01	0.01	0.95
Complete records	3.06	0.02	0.01	0.06	0.07	0.02	0.02	0.71	0.07	0.02	0.02	0.69
FCS‐MI	3.00	0.02	0.02	0.95	0.10	0.02	0.02	0.95	0.10	0.02	0.02	0.95
JM‐MI, unstransformed Y	3.00	0.02	0.02	0.96	0.10	0.02	0.02	0.95	0.10	0.02	0.02	0.95
JM‐MI, square root of Y	3.00	0.02	0.02	0.96	0.10	0.02	0.02	0.95	0.10	0.02	0.02	0.95
JM‐MI, log(Y)	3.00	0.02	0.02	0.96	0.10	0.02	0.02	0.95	0.10	0.02	0.02	0.96
Poisson – $β = (3, 0.3, 0.3)$
True value	3.00			0.95	0.30			0.95	0.30			0.95
Full data	3.00	0.01	0.01	0.95	0.30	0.01	0.01	0.95	0.30	0.01	0.01	0.95
Complete records	3.04	0.02	0.02	0.32	0.27	0.02	0.02	0.62	0.27	0.02	0.02	0.65
FCS‐MI	2.99	0.02	0.02	0.86	0.29	0.02	0.02	0.93	0.30	0.02	0.02	0.93
JM‐MI, unstransformed Y	2.99	0.02	0.02	0.90	0.29	0.02	0.02	0.92	0.31	0.02	0.02	0.93
JM‐MI, square root of Y	3.00	0.02	0.02	0.94	0.30	0.02	0.02	0.95	0.30	0.02	0.02	0.94
JM‐MI, log(Y)	3.01	0.02	0.02	0.92	0.30	0.02	0.02	0.97	0.29	0.02	0.02	0.95
Poisson – $β = (- 0.5, 0.1, 0.1)$
True value	−0.50			0.95	0.10			0.95	0.10			0.95
Full data	−0.51	0.08	0.08	0.95	0.10	0.07	0.08	0.95	0.10	0.07	0.07	0.96
Complete records	−0.38	0.10	0.10	0.73	0.09	0.10	0.10	0.94	0.09	0.10	0.09	0.96
FCS‐MI	−0.52	0.09	0.09	0.95	0.10	0.09	0.09	0.95	0.10	0.09	0.09	0.96
JM‐MI, unstransformed Y	−0.48	0.08	0.08	0.94	0.10	0.09	0.09	0.95	0.09	0.09	0.08	0.95
JM‐MI, square root of Y	−0.51	0.09	0.09	0.95	0.10	0.09	0.09	0.95	0.09	0.09	0.08	0.96
JM‐MI, log(Y)	−0.49	0.10	0.09	0.97	0.11	0.11	0.10	0.97	0.10	0.11	0.09	0.98
JM‐MI, categorical Y	−0.52	0.09	0.09	0.95	0.10	0.09	0.09	0.95	0.09	0.09	0.09	0.95
Count covariate
True value	0.30			0.95	0.30			0.95	0.30			0.95
Full data	0.31	0.12	0.12	0.95	0.30	0.03	0.04	0.94	0.30	0.06	0.06	0.95
Complete records	0.24	0.15	0.16	0.92	0.27	0.05	0.05	0.87	0.27	0.08	0.08	0.94
FCS‐MI	0.33	0.15	0.16	0.93	0.29	0.04	0.05	0.93	0.30	0.07	0.07	0.95
JM‐MI, unstransformed Y	0.32	0.15	0.15	0.94	0.30	0.04	0.05	0.94	0.30	0.07	0.07	0.95
JM‐MI, square root of Y	0.39	0.15	0.14	0.91	0.27	0.04	0.04	0.89	0.30	0.08	0.07	0.95
JM‐MI, log(Y)	0.98	0.14	0.10	0.00	0.06	0.04	0.02	0.00	0.29	0.08	0.07	0.96

Mean, model SE (mSE), empirical SE (eSE), and coverage level for each parameter. Data are MAR (see text) and we compare complete records (CR) to both FCS and JM imputation. For JM imputation we compare including Y either untransformed, square rooted, log‐transformed or as a categorical variable.