Table 1.

Results of hierarchical generalized linear mixed models (GLMMs) examining the contribution of different predictors for fixation selection.

		Centrality	Saliency	ROI	Saliency × ROI	Saliency × Valence	R 2
1	Centrality	0.554 [0.553, 0.555]					0.154 [0.153, 0.155]
2	+Saliency	0.266 [0.265, 0.268]	0.574 [0.572, 0.577]				0.252 [0.252, 0.253]
3	+ROI	0.288 [0.286, 0.289)	0.544 [0.542, 0.547]	0.506 [0.502, 0.509]			0.318 [0.317, 0.319]
4	+Saliency × ROI	0.287 [0.285–0.289)	0.526 [0.524–0.529)	0.509 [0.505, 0.512]	−0.103 [−0.107, −0.100]		0.318 [0.318, 0.319]
5	+Saliency × Valence	0.288 [0.286, 0.290]	0.528 [0.525, 0.530]	0.510 [0.507, 0.514]	−0.104 [−0.108, −0.100]	0.002 [−0.001, 0.004]	0.320 [0.319, 0.321]

Standardized regression weights and explained variance (R²) for models comprising an increasing number of predictors. Models are nested and include predictors in models shown above. All values were calculated by bootstrapping 100 sets of not-looked-at grid cells and performing GLMMs for each set. Estimates represent means of weights from each bootstrapping iteration. Values in brackets represent the 2.5^th and 97.5^th percentile rank as an unbiased estimate of the 95% confidence interval.