the percent of observed agreement between the twostimulus scenarios was very high for all behaviors, ranging from 64% to 100% (Table 2). However, for many behaviors, the percent of agreement was impacted by very low or very high prevalence. For example, ‘growl’ was exhibited by only 8.9% (four dogs) of dogs toward the real dog and by 15.6% (seven dogs) of dogs toward the fake dog but has a consistent result of 80% between the two conditions.