Table 3.

Information-theoretic model selection results of all considered negative-binomial generalized linear models for predicting the time (months) to onset of sleep-wake disturbances (Onset). Predictor variables were the age in years at initial TBI (Age), Gender, severity of initial TBI (Severity), and the total number of TBIs prior to the onset of sleep-wake disturbances (TBIs).

Model	K a	Log-Lik b	AICc c	ΔAICc d	Wt e
Onset ~ Age + Severity + TBIs	6	−492.15	996.70	0.00	0.50
Onset ~ Age + Severity + Gender + TBIs	7	−491.92	998.37	1.67	0.22
Onset ~ Age + TBIs	4	−495.33	998.83	2.13	0.17
Onset ~ Age + Gender + TBIs	5	−495.15	1000.57	3.87	0.07
Onset ~ Age	3	−498.61	1003.31	6.61	0.02
Onset ~ Age + Gender	4	−498.12	1004.42	7.72	0.01
Onset ~ Age + Severity	5	−498.07	1006.42	9.72	0.00
Onset ~ Age + Severity + Gender	6	−497.54	1007.47	10.77	0.00
Onset ~ TBIs	3	−502.17	1010.43	13.73	0.00
Onset ~ Severity + TBIs	5	−500.77	1011.80	15.10	0.00
Onset ~ Gender + TBIs	4	−501.99	1012.16	15.46	0.00
Onset ~ Severity + Gender + TBIs	6	−500.55	1013.49	16.79	0.00
Onset ~ 1	2	−505.54	1015.11	18.41	0.00
Onset ~ Gender	3	−505.07	1016.23	19.53	0.00
Onset ~ Severity	4	−505.50	1019.16	22.46	0.00
Onset ~ Severity + Gender	5	−505.01	1020.29	23.59	0.00

^a Number of model parameters. ^b log-likelihood of model. ^c Akaike’s Information Criterion (AIC) corrected for small sample size. ^d Difference between AIC_c of model and AIC_c of top-ranked model. ^e Model weight.