# A Method for Assessing Dogs in a Test Evaluating Dogs’ Suitability for Animal-Assisted Education

## TL;DR

This study proposes a new method for assessing dogs for animal-assisted education using objective behavioral ratings and AI-defined behaviors to improve consistency among raters.

## Contribution

A new approach using AI-defined behaviors to enhance inter-rater reliability in dog assessments for animal-assisted education.

## Key findings

- The BR method with AI definitions showed satisfactory inter-rater reliability in 7 out of 9 subtests.
- No significant differences were found between the BR method and the subjective rating method in subtest scores.
- AI-defined behaviors can be used to evaluate dogs based on temperament and momentary emotional states.

## Abstract

Objective behavioral rating (BR) is one of the methods for assessing dog behavior. In this approach, dogs are rated for behavior based on the observations made during a behavioral test. However, the degree of agreement among independent raters can differ. These variations could be attributed to different or unclear definitions of behavioral patterns. Thus, the aim of this study was to propose a new approach for assessing dogs’ suitability for animal-assisted education (AAE) with the use of the BR method and the definitions of dog behaviors proposed by artificial intelligence (AI). The analysis relied on video footage of dogs that participated in tests assessing the animals’ suitability for AAE. The results indicate that the BR method based on AI definitions produced satisfactory inter-rater reliability in a test evaluating dog behavior. A novel finding of this study is that the BR method can be used in dog assessments based on temperament traits and momentary emotional states and that dog behaviors can be evaluated based on the definitions proposed by AI. Thus, we conclude that the proposed approach gives promising outcomes and can be used to develop new tests for assessing dogs’ suitability for AAE.

In objective behavioral rating (BR), dogs are rated for behavior based on the observations made during a behavioral test. However, behavioral definitions can differ between raters, and the definitions proposed by artificial intelligence (AI) can help standardize the applied measures. The aim of this study was to propose a new approach for assessing dogs’ suitability for animal-assisted education (AAE) with the use of the BR method and the definitions of dog behaviors proposed by AI. The analysis relied on video footage of 25 dogs that participated in tests assessing the animals’ suitability for AAE. The dogs were rated by five independent observers. Inter-rater reliability was satisfactory in 7 out of 9 subtests (0.715–0.856) and low in the remaining 2 subtests (0.667 in Subtest 2 and 0.617 in Subtest 5). The results indicate that the BR method based on AI definitions produced satisfactory inter-rater reliability in a test evaluating dog behavior. A comparison of subtest scores in the BR method and the subjective rating method did not reveal significant differences. Thus, it can be concluded that the proposed approach gives promising outcomes and can be used to develop new tests for assessing dogs’ suitability for AAE and, perhaps, other types of work.