Table 1.
Definition of the variables coded in the tests and their inter-observer reliability [Cohen's kappa or intraclass correlation (ICC)].
| Subtest, variable | Definition | Data processing |
|---|---|---|
| 1. Interaction with the caretaker | ||
| Reaction to petting | When C tried to pet the dog, the dog: 0, avoided contact (turned or moved away); 1, passively tolerated contact (no sign of contact seeking or avoidance); 2, showed a little contact seeking (shortly sniffed C, kept eye contact); 3, actively sought contact with C (cuddle up, lick, and climb in lap). If the dog behaved differently at the beginning vs. the end of the trial, the mean of the scores assigned to the two behaviors was given. | Coded separately for the four trials. For the analysis, the mean of the four trials was calculated. ICC = 0.927, F14,14 = 13.792, p < 0.001 |
| Latency of approach | From the moment C stepped outside the dog's reach until the dog got within arms' reach of C. If the dog never stepped out of reach, the latency was 0; if the dog did not approach C, the maximum (15 s) was given. | Coded separately for the three approaches. For the analysis, the raw latency was recoded into categories based on its histogram: 0, 0–0.9 s; 1, 1–2.3 s; 2, 2.4–4.5 s; 3, 4.5–9 s; 4, 9–13 s; 5, 13–15 s. Then the mean score of the three approaches was calculated. ICC = 0.816, F14,14 = 9.863, p < 0.001 |
| 2. Food motivation and 6. Directional gesturing | ||
| Accept food | If the dog ate food (1) or not (0). | Coded separately for the two tests. For the analysis, the two variables were summed. Cohen's kappa = 0.694, N = 15, p < 0.001 |
| 3. Greeting by an unfamiliar experimenter | ||
| Reaction to petting | When E tried to pet the dog, the dog: 0, avoided contact (turned or moved away); 1, passively tolerated contact (no sign of contact seeking or avoidance); 2, showed a little contact seeking (shortly sniffed E, kept eye contact); 3, actively sought contact with E (cuddle up, lick, and climb in lap). If the dog behaved differently at the beginning vs. the end of the trial, the mean of the scores assigned to the two behaviors was given. | Coded separately for the three trials. For the analysis, the mean of the three trials was calculated. ICC = 0.683, F14,14 = 3.020, p = 0.024 |
| Latency of approach | From the moment E called the dog/stepped outside the dog's reach until the dog got within arms' reach of E. If the dog approached E before her call or never stepped out of reach, the latency was 0; if the dog did not approach E, the maximum (15 s) was given. | Coded separately for the three trials. For the analysis, the raw latency data were recoded into categories based on its histogram: 0, 0–0.9 s; 1, 1–2.3 s; 2, 2.4–4.5 s; 3, 4.5–9 s; 4, 9–13 s; 5, 13–15 s. Then the mean score of the three trials was calculated. ICC = 0.973, F14,14 = 47.054, p < 0.001 |
| 4. Training for eye contact | ||
| Frequency of eye contacts | The number of eye contacts the dog established during the test phase (3 min). If the dog did not pass the pretraining phase, 0 (the minimum) was given. | For the analysis, the raw frequency data were recoded into categories based on its histogram: 0, 0–1; 1, 2–10; 2, 11–21; 3, >21. Cohen's kappa = 1.000, N = 15, p < 0.001 |
| 5. Potentially threatening moving object | ||
| Type of reaction | The object stopped because the dog: 0, moved away, in the opposite direction as C; 1, moved behind C; 2, was passive; 3, approached the object. | Coded separately for the two trials. For the analysis, the mean of the two trials was calculated. ICC = 0.997, F14,14 = 341.714, p < 0.001 |
| Distance from object | How far the object was from the dog when it stopped: Score 0, ≥4 m; Score 1, ≥2 and <4 m; Score 2, ≥1 and <2 m; Score 3, <1 m. | Coded separately for the two trials. For the analysis, the mean of the two trials was calculated. ICC = 0.995, F14,14 = 200.714, p < 0.001 |
| Latency of sniffing | From the moment C called the dog to the object until the dog approached it at <10 cm. If the dog approached the object on its own while it was still moving, the latency was 0; if the dog did not approach the object at all, the maximum (30 s) was given. | For the analysis, the raw latency data were recoded into categories based on its histogram: 0, 0 s; 1, 1–5 s; 2, 5–20 s; 3, 20–30 s. Cohen's kappa = 1.000, N = 15, p < 0.001 |
| 6. Directional gesturing | ||
| Choice | The plate the dog approached at <10 cm: 0, none; 1, any plate. | Coded separately for the three trials. For the analysis, the number of valid choices out of three was calculated. Cohen's kappa = 1.000, N = 15, p < 0.001 |
| 7. Emotion recognition | ||
| Choice | The object the dog approached at <10 cm: 0, none; 1, any object. | Coded separately for the two trials. For the analysis, the number of valid choices out of two was calculated. Cohen's kappa = 1.000, N = 15, p < 0.001 |
| 8. Threatening approach | ||
| Type of reaction | The dog's final reaction (when the test was terminated): 0, active avoidance (moved away or behind C); 1, passive (no movement toward or away from E); 2, ambivalent (few hesitant steps toward/away from E, may show tail wagging); 3, friendly/appeasing (approached E). | Remained the same for the analysis. Cohen's kappa = 1.000, N = 15, p < 0.001 |
| Distance from E | How far E was from the dog when she terminated the approach: Score 0, >2 m; Score 1, 1–2 m; Score 2, <1 m; Score 3, the dog approached E. | Remained the same for the analysis. Cohen's kappa = 1.000, N = 15, p < 0.001 |
E, experimenter; C, caretaker.