In 2005 the Swedish Armed Forces (SAF) started to build their own breeding program. They decided on a more or less closed population of German Shepherds of approximately 80 bitches and 15 males, all owned by SAF, producing around 200-300 puppies per year. Both working dogs and replacement breeders are recruited from the dogs produced. When a dog born within the program is approximately 16 months old, it is given a standardized temperament test developed by the SAF. The temperament test is in the form of a test battery containing 12 sub-tests. Based on test results and health data it is decided whether the dog should be rejected or if it should be used for breeding or put into training. The dog’s behaviour during the test is recorded using two protocols simultaneously; in the first protocol the rating method is termed “behavioural ratings” (BR), in the second “subjective ratings” (SR).
Our aim is to study how well the SAF temperament test measures genetic differences between dogs, both the BR and the SR method. We also want to try to identify characteristics of the test, as well as of the statistical treatment of ratings, that can be altered to improve the test as a tool for selection of breeders.