A Bayesian hierarchical logistic regression model of multiple informant family health histories

Table 1 Model fit and classification accuracy of five candidate models from the Hierarchical Bayesian Logistic Regression of MIFHH of Type 2 Diabetes

Terms	Family Level-2			Informant Level-2
Terms	Pars	DIC	AUC	Pars	DIC	AUC
Null model	3	2508.33	–	3	2508.33	–
Level-1 and Level-2 Intercepts Only	47	2411.2	0.636	130	2465.03	0.694
M1 + Degree of relation^a + Same gender^a	49	2388.91	0.671	132	2457.72	0.708
M2 + Informant gender^b + Informant is obese^b	147	2394.01	0.686	396	2400.52	0.714
M2 + Smokes^a + Uses alcohol^a + Healthy weight^a	52	2227.54	0.747	135	2263.68	0.77
M3 + M4	150	2224.39	0.762	399	2237.7	0.78

ISSN: 1471-2288