Table 5. Results of GLMM testing variation in future direction changes.

Given are results for the full model, including all interactions, and of the final model, from which non-significant interaction terms were removed. Significance of interpretable terms in the final model (interaction terms and main effects of terms not included in an interaction) was assessed with likelihood ratio tests.

	Full model		Final model		LRT
	β ± se	z	β ± se	z	χ21	p
Intercept	−1.80 ± 0.38	−4.72	−1.76 ± 0.36	−4.85
Subgroup size	−0.30 ± 0.33	−0.91	−0.43 ± 0.25	−1.69
Focal grunt (yes)	0.26 ± 0.52	0.51	0.46 ± 0.48	0.95	0.89	0.3444
Focal other vocalization (yes)	0.79 ± 0.72	1.10	0.50 ± 0.63	0.79	0.61	0.4367
Subgroup grunt	−0.03 ± 0.28	−0.11	−0.10 ± 0.24	−0.40	0.16	0.6917
Subgroup other	−0.40 ± 0.27	−1.48	−0.32 ± 0.25	−1.30
Subgroup twitter	0.51 ± 0.30	1.69	0.49 ± 0.25	1.94	3.86	0.0493
Number of associated species	−0.21 ± 0.28	−0.75	−0.31 ± 0.25	−1.22	1.48	0.2238
Vocal rate of other species	0.15 ± 0.28	0.54	−0.03 ± 0.23	−0.12	0.01	0.9064
Sex (male)	0.53 ± 0.63	0.84	0.71 ± 0.61	1.17	1.36	0.2437
Control change point	1.19 ± 0.51	2.33	1.31 ± 0.48	2.76	7.50	0.0062
IA subgroup size: focal grunt	−1.25 ± 0.68	−1.84
IA subgroup size: focal other	1.46 ± 0.93	1.58
IA subgroup size: subgroup grunt	0.17 ± 0.28	0.61
IA subgroup size: subgroup other	−0.79 ± 0.34	−2.28	−0.56 ± 0.29	−1.94	4.44	0.0352
IA subgroup size: subgroup twitter	0.04 ± 0.33	0.11
IA associated species: vocal rate of other species	0.33 ± 0.21	1.57

Note:

Reference levels of categorical variables are: focal grunt = ‘no,’ focal other = ‘no,’ and sex = ‘female.’ LRT, likelihood ratio test, IA, interaction.