Validation of a short odour discrimination test for working dogs

A short odour discrimination test has been designed to allow rapid quality assurance of odour recognition by detection dogs. The test comprises five repeats per target and a minimum of 20 associated non-target odours. The mean time taken to conduct the test is 5.6 min per target type. A pass criterion of “a detection rate at least 70% greater than false alarm (FA) rate, with a 15% cap on total allowable false alarms” is used which equates to 4/5 correct indications and 2 FAs, or 5/5 correct indications and 3 FAs; the probability of passing this test by chance is <1%. A Microsoft Excel™ programme has been written to rapidly generate balanced running orders that allow search runs to be truncated following correct indications; this speeds up testing whilst maintaining standardisation; the programme is available free-to-use. The test's internal validity has been measured by conducting test re-test analysis on a range of target types on 19 operational search dogs, and external validity has been measured by completing the test and an equivalent operationally relevant building search on 26 operational search dogs. In both cases there is good overall reliability (kappa ≥ 0.80). The test is thus deemed suitable for complementary assessment of detection dog ability during detailed accreditation procedures or as a standalone quality assurance test in between accreditation or licensing.