Table 1.

Goodness of Fit Criteria, Model Adequacy and Proportional Placement for 1-Class to 6-Class Models

Goodness of Fit Criteria									Average Posterior Probability						Relative Entropy	Percent in Each Class
Class	Order	n	LL	LMR-LR	LMR-LR P	AIC	BIC	Δ BIC	Class 1	Class 2	Class 3	Class 4	Class 5	Class 6		Class 1	Class 2	Class 3	Class 4	Class 5	Class 6
1	Linear	5 976	−114 754.4	—	—	−114 756.4	−114 766.48	—	100.00	—	—	—	—	—	—	100.00	—	—	—	—	—
2	Quadratic	5 976	−79 992.71	66 956.056	p < .001	−79 999.708	−80 035.013	34 731.5	0.99	0.98	—	—	—	—	0.96	63.52	36.48	—	—	—	—
3	Quadratic	5 976	−74 435.97	10 703.13	p < .001	−74 446.97	−74 502.449	5 532.6	0.97	0.94	0.97	—	—	—	0.92	50.57	26.94	22.49	—	—	—
4	Quadratic	5 976	−72 938.16	2 885.012	p < .001	−72 953.155	−73 028.808	1 473.6	0.94	0.88	0.91	0.95	—	—	0.87	39.90	24.55	18.48	17.07	—	—
5	Quadratic	5 976	−71 613.98	2 550.63	p < .001	−71 632.977	−71728.804	1 300.0	0.85	0.93	0.85	0.87	0.93	—	0.85	17.97	39.63	12.78	12.85	16.77	—
6	Quadratic	5 976	−70 777.17	1 611.816	p < .001	−70 800.167	−70 916.168	812.6	0.85	0.90	0.84	0.86	0.90	0.82	0.83	10.44	35.57	11.67	10.90	13.57	17.85

Notes: Change in the Bayesian Information Criterion (BIC) measures the weight of evidence against the previous, less complex model. The best-fit model is related to the smallest negative value. AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; ΔBIC: Change in BIC from the previous model; LL = Log-likelihood; LMR-LR = Lo-Mendell-Rubin likelihood ratio test statistic; LMR-LR P = p-value for the Lo-Mendell-Rubin test; n = Sample size.

Average posterior probability is a measure of the internal reliability for each trajectory class (values closer to 1.0 are ideal).

Entropy is the ability of the model to provide well-separated classes (closer to 1.0 is ideal). A value greater than 0.80 indicates less classification uncertainty.

Bold indicates chosen 5-class solution.