Combining estimators of a common parameter across samples

Figures & data

Table 1. Estimated parameters and SE's from two models, with coefficients for Trt and for Trt*bonCtr, in separate study centres. ‘*’ denotes structural 0.

		Model 1		Model 2
j
1	21	0.307	0.166	0.383	0.154	1.083	0.420
2	17	0.266	0.205	0.266	0.205	*	*
3	18	0.682	0.205	0.687	0.207	−0.187	0.427
4	27	0.400	0.343	0.406	0.340	0.243	0.840
5	46	−0.159	0.183	−0.147	0.177	0.367	0.452
6	31	0.407	0.294	0.410	0.295	−0.186	0.620
7	17	−0.246	0.293	−0.246	0.293	*	*
8	59	−0.035	0.154	−0.031	0.150	0.288	0.313
9	56	0.415	0.181	0.424	0.177	−0.304	0.413
10	31	0.646	0.296	0.645	0.296	−0.012	0.571
11	22	0.591	0.250	0.603	0.216	1.737	0.623
12	39	0.353	0.286	0.342	0.281	−0.313	0.623
13	27	0.181	0.202	0.188	0.187	−0.437	0.362
14	53	−0.269	0.211	−0.296	0.203	0.671	0.365
15	17	0.352	0.655	0.526	0.626	−1.365	1.198
16	42	0.026	0.160	0.033	0.162	−0.216	0.386
17	23	−0.107	0.400	−0.122	0.388	0.896	0.710
18	24	−0.219	0.312	−0.221	0.299	0.543	0.489

Table 2. Information on β in combined and separate skew-normal samples. In all cases, c=0.5.

λ	0.25	0.5	1.0	.05	0.1	0.25	0.5	1.0	2.0
σ	1	1	1	2	2	2	2	2	2
	.0382	.1378	.4162	.0638	.0676	.0929	.1718	.3868	.8053
	.0377	.1307	.3572	.0016	.0063	.0377	.1307	.3572	.7345
	.0002	.0034	.0389	.0000	.0000	.0001	.0008	.0097	.0563