Table 1.

Percentage of total deviance explained by the logit-linear and non-linear regression splines in Fig. 1.^a

	Deviance explained (%)
Source	A	B	C	D
Linear	97.242	96.40	99.438	99.896
Non-linear 2–3	2.732	2.32	0.328	0.029
Non-linear 4–5	0.026	1.28	0.234	0.075

^aThe rows of the table show the percentage of the total deviance explained by the logit-linear curve and the additional deviance explained by logit splines with 3 or 5 df. The second row gives the additional deviance explained by df$\hspace{0.17em}=3$ over df$\hspace{0.17em}=1$ and the third row the additional deviance explained by df$\hspace{0.17em}=5$ over df$\hspace{0.17em}=3$. The logit-linear curve explains >96% of the deviance for all datasets.