Generating social network data using partially described networks: an example informing avian influenza control in the British poultry industry

Table 1 Crude odds ratios before and after removal of records with missing data: scenario 1 analyses

	Full data (n = 662 farms)			Following the removal of records (n = 348 farms)
Farm-level predictors	OR^a	s.e. OR^b	p-value	OR^a	s.e. OR^b	p-value	% change
L vs. S bird count^e	0.284	1.292	<0.0001	0.185	1.367	<0.0001	34.9^c
L vs. S house count^e	0.354	1.306	<0.0001	0.299	1.327	<0.0001	15.5
Indoors	0.298	1.289	<0.0001	0.301	1.307	<0.0001	1.0
Free-range	5.266	1.298	<0.0001	5.010	1.317	<0.0001	4.9
Housing other	0.829	1.510	0.650	0.598	1.590	0.268	27.9^c
Partial housing	1.009	1.941	0.990	0.763	2.209	0.733	24.4
Integrated	0.271	1.222	<0.0001	0.258	1.309	<0.0001	4.8
East	1.905	1.298	0.014	2.545	1.392	0.005	25.1
Scotland	0.333	2.144	0.150	0.308	2.899	0.268	7.5
Wales	0.119	2.087	0.004	0.192	2.848	0.115	38.0^d
West	0.418	1.287	0.001	0.401	1.429	0.010	4.1

^aOR = odds ratio; ^bs.e. = standard error of the odds ratio; ^c >25% change in odds ratio but direction of association and significance is comparable; ^dsingle variable for which there is >25% change in odds ratio and no change in direction of association, but significance is altered; ^eL = large, S = small.

ISSN: 1746-6148