Development of a metric for tracking and comparing population health based on the minimal generic set of domains of functioning and health

Table 4 Results on predictive validity – comparison of model-fit criteria for four different models

	Adjusted R-square	Percentage of deviance explained	AIC
Model 1 including only covariates	15.2	23.6	3362
Model 2 including covariates and the health metric	17.5	26.2	3251
Model 3 including covariates and the general health question	16.7	25.5	3285
Model 4 including covariates, the general health question and the health metric	17.9	26.6	3240

For wave-4 data, four different additive logit-models predicting mortality in 2008 to 2012 are compared based on three model-fit criteria: adjusted R-square, percentage of deviance explained, and Akaike Information Criterion (AIC). Covariates considered in all four models include sex, age, education, income, and health conditions. To permit a fair comparison of criteria, the same subset of data with complete responses in all the variables considered over the four models was used. Where included, age and the health metric are modeled in a flexible, non-parametric way using P-splines

ISSN: 1478-7954