Table 2.

General linear (mixed) models testing for effects of dummy type and the relative bar width of focal fish on the behavior of male and female focal fish

(a) Experiment 1: Charges and displaysa
	Females (n = 16)				Males (n = 17)
	β	SE	t	P	β	SE	t	P
Dummy type (wide-barred)	− 0.530	0.671	− 0.789	0.443	− 1.098	0.478	− 2.295	0.037
Relative bar widthb	0.044	0.642	0.069	0.946	0.076	0.471	0.161	0.874
First dummy presentation (yes)	0.778	0.671	1.158	0.266	2.078	0.478	4.343	0.0006

(b) Experiment 2: Latency time to explorationc
	Females (n = 34)				Males (n = 27)
Relative bar widthb	− 1.590	0.766	− 2.076	0.046	− 1.057	0.820	− 1.289	0.209

(c) Experiment 2: Latency time to intrusiond
	Females (n = 15)				Males (n = 15)
Dummy type (wide-barred)	− 0.109	0.224	− 0.488	0.633	0.279	0.275	1.014	0.328
Relative bar widthb	0.393	0.255	1.542	0.147	− 0.351	0.258	− 1.359	0.197

(a) Charges and displays of focal fish against wide-barred and narrow-barred dummies, (b) latency time to exploration of an unfamiliar area of the tank (without dummy presentation), (c) latency time to intrude into a territory guarded by a wide-barred or a narrow-barred dummy. Effect estimates (β), standard error (SE), t-value and P value are shown for each of the fixed factors included in each respective model. Significant P values are indicated in bold

^aThe counts for each behavior (frontal displays, lateral displays, swimming with dummy and charges towards dummy) were normalized using their medians and interquartile ranges, and then summed up and collectively referred to as charges and displays

^bRelative bar width was scaled and centered

^cSquare-root transformed

^dLog transformed