Table 4. Final model predicting individual look durations.

	Fixed Effects					Random Effects
Model Parameters	Β	SE	t	df	p	Variance	SD	χ2	df	p
Intercept	0.83	0.02	53.89	126	< .001	0.02	0.16	866.46	126	< .001
Log10 Look Count	-0.51	0.03	-14.94	126	< .001	0.07	0.27	314.48	126	< .001
Lag 1 Duration	0.05	0.02	2.50	126	.014	0.01	0.08	150.48	126	.068
Model Fit Statistics	Deviance	Number of Parameters		Level 1 Observations (Individual Looks)		χ2	p	AIC		BIC
	2546.53	10		3409		100.66	< .001	2566.53		2581.86

Individual look duration is uniquely predicted by log10 look count (habituation) and lag 1 duration (temporal dependency). The χ² describes a comparison to a model with no temporal dependency (Lag 1) effects (see text). The inclusion of Log10 Look Count as a predictor accounted for 27.0% more variance than a model with the intercept alone. The inclusion of Lag 1 Duration as a predictor accounted for 0.6% more variance than a model with Log10 Look Count. For the model’s equation and unstructured (full) covariance matrix see S1 Text.