We coded the presence of the head instead of the entire body because the dogs often straddled 2 areas; to avoid confusion when coding data, only the presence of the head was considered. Videos from 15 dogs were coded both by the experimenter and by a naïve observer to measure interrater reliability. Pearson correlation between the 2 raters for presence in areas 1-4 was 0.981 (P < 0.001), 0.567 (P ¼ 0.028), 0.989 (P < 0.001), and 0.99 (P < 0.001), respectively. The relatively low correlation in area 2 was because of 1 dog that straddled areas 1 and 2 for much of the coding period and whose actual placement was difficult to determine on the video. When correlations were analyzed without this dog, Pearson correlation for area 2 became 0.891 (P < 0.001).