Table 5.

Linear mixed-effects model of the effects of restraint types and the order of restraint types used on canine behaviours when testing with food treats.

Food treat
	Lip-licking	Looking at experimenter	Panting	Ear-flatten	Tail-high
Restraint type (harness)	–	β: 0.087 SE: 0.041 p: 0.039	–	–	–
Order of restraint	β: −0.073 SE: 0.031 p: 0.021	–	–	–	–
Order 1	Median: 0.19 IQR: 0.19	Median: 0.15 IQR: 0.28	Median: 0.72 IQR: 0.79	Median: <0.01 IQR: 0.34	Median: <0.01 IQR: <0.01
Order 2	Median: 0.1 IQR: 0.2	Median: 0.11 IQR: 0.29	Median: 0.73 IQR: 1	Median: <0.01 IQR: 0.31	Median: <0.01 IQR: <0.01

Restraint type: collar was used for comparison. Order of restraint: the order of restraint types used was randomly determined. The order of each dog participating in the study was also entered into the model but was excluded by the backward elimination process. Lip-licking: numbers of lip-licking observed per second (analysed after transformation to the power of 0.8). Looking at experimenter: proportion of time looking at the experimenter (analysed after transformation to the power of 0.5). Panting: proportion of time panting. Ear-flatten: proportion of time the dog kept its ears flatten. Tail-high: proportion of time the dog kept its tail in a high position. IQR, interquartile range. Paw-lifting, body shaking, sniffing, and vocalisation were seldomly observed. –: the predictor was excluded from the model due to the backward elimination process.

In the mixed-effect models, dogs displayed a significantly lower frequency of lip-licking behaviour in the second trial when tested with food treats (p = 0.021) and the toy (p=0.048) (Table 5 and Appendix Table 2). Finally, when testing with food treats, there was a significantly higher frequency of lip-licking behaviour during the first trial utilising the harness (p = 0.018) (Appendix Table 3) than during the second.