Table 2. Forward selection of the best model.

$H_0$	$H_1$	$-2\log \lambda \left(S\right)$	P-value
ISO	IM1	603.14	1.147E−262
IM1	IIM1	413.12	7.673E−92
IIM1	IIM2	340.44	1.187E−74

Results refer to the data of Wang and Hey (2010).