Table 1.

Comparison of Quasi-likelihood under Independence Criterion (QIC) parameters for model selection between the Wind model and the Wind+SP model

Model name	Variables	K	QIC	QICω	Χ2	C-index
Females
Wind+SP	SP * windDir + SP:windSp + windDir:windSp	5	6336.97	0.99	92.92	0.61
Wind	windDir * windSp	2	6360.42	< 0.001	62.12	0.56
Males
Wind+SP	SP * windDir + SP:windSp + windDir:windSp	5	4729.39	0.99	71.94	0.61
Wind	windDir * windSp	2	4770.18	< 0.001	17.69	0.53

K = number of parameters; QICω = QIC weight, assesses weight of evidence in support of model; Χ² = compares log-likelihood of the full model to the log-likelihood of a null model; C-index = concordance index, measures agreement between model predictions and observed response. Smaller QIC and higher QICω indicate better support for the model; larger Χ² values indicate a better model fit. Parameters joined by ‘:’ indicate an interaction; parameters joined by ‘*’ indicate an interaction plus the main effects of the singular variables. As wind speed does not vary within decision points, it could not be fitted as a fixed effect (SI Appendix, Supplementary Materials). The model name indicates how the model is referred in the text. Models are ordered by QICω.

Notes – SP = sound pressure, pascals; windDir = wind direction relative to segment bearing, °; windSp = wind speed, m/s.