Table 1.

AIC, number of parameters, and minus twice the maximum value of the log-likelihood function (-2LL) of the various models fitted to the ELSA data

Hazard distributions	Initial distribution	Binomial distribution(s)	#Parameters	-2LL	AIC
$\mathcal{E}\mathcal{E}\mathcal{E}$	Age	Standard univariate	9	31124	31142
$\mathcal{E}\mathcal{E}\mathcal{E}$	Age	Extended univariate	11	31026	31048
$\mathcal{E}\mathcal{E}\mathcal{E}$	Age	Bivariate	12	29708	29732
$\mathcal{G}\mathcal{G}\mathcal{G}$	Age	Bivariate	15	29616	29646
$\mathcal{G}\mathcal{G}\mathcal{G}+\text{gender}$	Age	Bivariate	18	29603	29639
$\mathcal{G}\mathcal{G}\mathcal{G}+\text{gender}$	Age $+$ gender	Bivariate	19	29598	29636
$\mathcal{W}\mathcal{W}\mathcal{W}$	Age	Bivariate	15	29621	29651
$\mathcal{G}\mathcal{W}\mathcal{W}$	Age	Bivariate	15	29619	29649
$\mathcal{G}\mathcal{W}\mathcal{W}+\text{gender}$	Age	Bivariate	18	29611	29647
$\mathcal{G}\mathcal{W}\mathcal{W}+\text{gender}$	Age $+$ gender	Bivariate	19	29606	29644
$\mathcal{W}\mathcal{G}\mathcal{G}$	Age	Bivariate	15	29688	29718

Letters for the hazard models ($\mathcal{E}$xponential, $\mathcal{G}$ompertz, and $\mathcal{W}$eibull) for $1\to 2$, $1\to 3$, and $2\to 3$, respectively