Table 6. The statistical output of the final Cox Proportional Hazards regression model.

Term	Regression Coefficient	Standard Error	DF	p-value
Cue P1	−0.816	0.123	1	<0.001
Cue P2	−1.026	0.133	1	<0.001
Cue P3	−0.816	0.123	1	<0.001
Cue P4	−1.170	0.137	1	<0.001
Cue P5	−1.466	0.147	1	<0.001
Cue P6	−1.287	0.139	1	<0.001
Cue P7	−1.107	0.128	1	<0.001
Cue P8	−1.350	0.141	1	<0.001
Cue P9	−1.749	0.158	1	<0.001
Cue Water	−1.352	0.064	1	<0.001
Test 2	−0.007	0.059	1	0.91
Test 3	−0.213	0.059	1	<0.001
Frailty (Dog)			18.6	<0.001

Includes all dogs that complete the cognitive bias testing (n = 20). These data describe the difference between the latency of dogs touching the target after the milk tone (reference condition) to each probe tone (CueP1–CueP9) and the water tone (Cue Water). Negative regression coefficients show a reduction in the likelihood of reaching a certain event, in this case, touching the target. Thus, the likelihood of touching the target is significantly less after probe and water tones than after milk tones. The risk of touching the target was not significantly different between test 1 and 2, but was significantly less in test 3 than test 1, indicating a reduced likelihood of touching the target over successive tests. The frailty term (“Dog”) refers to the dog being tested, which is treated in this model as a random effect due to repeated measures on each dog. The term “Dog” also had a significant effect on likelihood of touching the target, meaning that individuals varied significantly in their latency to touch the target.