Table 3.

Model summary and fit statistics for conditional models (including covariates) in the exploratory data. Italicized fit statistics indicate models that obtained information criteria, but may have untrustworthy standard errors due to non-positive definite 1^st-order product derivative matrix. The problem parameter is indicated. Blank entries indicate models with non-positive definite Fisher Information matrix, for which information criteria cannot be computed. The two best BIC values are bolded.

Model	n class	Params	Log-likelihood	AIC	BIC	aBIC	Fisher Info. Error?	1st-order Deriv. Error? (Bad param.)
m1 (conditional)	2	12	−1917.08	3858.15	3910.91	3872.82	No	No
m2 (conditional)	3	19	−1799.51	3637.01	3720.55	3660.23	No	No
m3 (conditional)	4	26	−1763.91	3579.82	3694.14	3611.60	No	No
m4 (conditional)	5	33	−1748.29	3562.58	3707.68	3602.91	No	No
m5 (conditional)	2	14	−1815.94	3659.88	3721.43	3676.99	No	Yes (class 1 on X3)
m6 (conditional)	3	22	−1763.59	3571.18	3667.91	3598.06	No	Yes (class 1 on X3)
m7 (conditional)	4	30	−1743.59	3547.18	3679.09	3583.84	No	No
m8 (conditional)	2	18	−1804.74	3645.48	3724.63	3667.48	No	No
m9 (conditional)	3	28	−1752.94	3561.89	3685.00	3596.11	No	Yes (class 2 on X3)
m10 (conditional)	4	38	−1738.58	3553.16	3720.24	3599.61	No	No
m11 (conditional)	2	24	−1801.65	3651.30	3756.82	3680.63	No	Yes (class 1 on X3)
m12 (conditional)	3						Yes
m13 (conditional)	2	20	−1743.22	3526.43	3614.37	3550.88	No	No
m14 (conditional)	3	31	−1731.19	3524.37	3660.68	3562.26	No	No
m15 (conditional)	4	42	−1721.62	3527.24	3711.91	3578.57	No	No
m16 (conditional)	2	24	−1741.37	3530.74	3636.26	3560.07	No	No
m17 (conditional)	3	37	−1727.67	3529.35	3692.03	3574.57	No	No
m18 (conditional)	4	50	−1716.44	3532.89	3752.73	3593.99	No	Yes (slope variance)
m19 (conditional)	2	30	−1740.44	3540.88	3672.79	3577.54	No	Yes (quadratic variance)
m20 (conditional)	3	46					Yes

Note. param. = parameter; AIC = Akaike information criterion; BIC = Bayesian information criterion; aBIC = sample-size adjusted BIC.