TABLE 3.

Mixed effect linear regression model on empirical log error ratio.

L1 Group	Estimate	Std. error	t	p
(Intercept)	0.08	0.04	1.86	0.065
Talker	0.31	0.04	6.92	<0.001
Window	0.28	0.04	6.26	<0.001
Talker:Window:Frequency	0.11	0.05	2.39	018
Pairwise comparisons
Multi-talker – Single-talker	0.62	0.10	5.87	<0.001
Early (Gates 2-3) – Late (Gates 4-7)	0.56	0.09	5.70	<0.001
Early Multi (F−) – Early Single (F−)	0.24	0.18	1.31	0.893
Early Multi (F+) – Early Single (F+)	0.93	0.18	5.17	<0.001
Late Multi (F−) – Late Single (F-)	0.74	0.18	4.13	0.002
Late Multi (F+) – Late Single (F+)	0.57	0.18	3.21	0.036
L2 Group
(Intercept)	0.10	0.04	2.49	0.015
Talker	0.48	0.04	−4.59	<0.001
Frequency	−0.14	0.03	11.35	<0.001
Pairwise comparisons
Multi-talker – Single-talker	0.96	0.08	11.24	<0.001
(F−) – (F+)	0.28	0.06	4.28	<0.001

L1 model: lmer[log ratio ∼ talker + window + talker:window:frequency + (talker| subject)]. L2 model: lmer[log ratio ∼ talker + frequency + (window| subject)].