Investigating behavior assessment instruments to predict aggression in dogs

• Published in: Applied animal behaviour science 2012-11, Vol.141 (3-4), p.139-148
• Main Authors: Bennett, Sara L., Litster, Annette, Weng, Hsin-Yi, Walker, Sheryl L., Luescher, Andrew U.
• Format: Article
• Language: English
• Subjects: Aggression; Behavior assessment instrument; behavior modification; Dogs; fearfulness; odds ratio; safety assessment; Shelter; temperament; Temperament test
• ISSN: 0168-1591; 1872-9045
• DOI: 10.1016/j.applanim.2012.08.005

This masked controlled study evaluated a group of dogs to determine if the results of two behavior assessments detected aggression in dogs that had a history of aggression according to a validated questionnaire for measuring behavior and temperament traits in dogs. Groups of dogs with or without a history of aggression were identified from owner-completed questionnaires for 67 dogs. Any dogs that had a maximum score of no greater than 1 for any question comprising aggression factors were placed in the low/no aggression group and any dogs that had a maximum score of 2 or higher on any question comprising the aggression factors were placed in the moderate to severe aggression group. This second group was further divided to separate moderate aggression from severe aggression. Two behavior assessments, Meet Your Match (MYM)™ Safety Assessment for Evaluating Rehoming™ (SAFER™) (SAFER) and a modified version of Assess-A-Pet (mAAP), were administered to each dog in random order by assistants masked to the dogs’ behavioral histories. The scores for each assessment were divided into binary categorizations (no aggression or aggression). For SAFER, the aggression category was further divided, separating dogs that showed fear, arousal or inhibited aggression from those that showed moderate aggression, and from those that showed severe aggression. The previously established categories for the mAAP of ‘no issue’, ‘unsocial’, ‘borderline’ and ‘fail’ were also used. Subtest scores for each assessment were also summed. With binary categorization, SAFER showed both lower sensitivity and specificity at 0.60 (95% confidence limits (CL)=0.44, 0.74) and 0.50 (95% CL=0.28, 0.72) respectively, than mAAP at 0.73 (95% CL=0.58, 0.85) and 0.59 (95% CL=0.36, 0.79) respectively. The odds ratio showed that an aggressive dog was 4.1-fold more likely to be classified in an aggression group by the mAAP test and 1.5 times more likely by SAFER. When the assessments were split into multiple categories, SAFER results were no longer significant, but mAAP maintained a statistically significant but weak correlation of 0.34 (P=0.005) with historical aggression categories. SAFER testing was unable to identify dogs with moderate aggression that could potentially be addressed with behavior modification. By independently selecting acceptable levels of false positive or false negative results for the assessment, summed score results could be used in shelters as an aid to selecting dogs for adoption. Be