Abstract/Summary

Fear is an important welfare problem for farm animals, including cattle. A variety of methods of assessing fear have been proposed, but the reliability and validity of these methods, and ways of improving these characteristics, have received little study. We conducted a series of experiments to assess the consistency of dairy calves’ responses of novel objects and to humans, and to investigate factors that might improve reliability. In the first experiment, latency to touch a novel object had moderate reliability (rs = 0.54), and latency to touch a stationary, familiar human had negligible reliability (rs = 0.26). Experiment 2a used the same test protocols, but with a shorter interval between repeat testing and using different stimuli in the two novel object tests; this change did not improve reliability (e.g. rs = 0.29 for the novel-object test). Reliability for this test was improved (rs = 0.58) in Experiment 2b, when the same object was used in both tests rather than a truly novel object being used the second time. Experiment 2a found ceiling effects in the response to human test associated with the short period during which approach responses were recorded. High reliability was found in Experiment 2b, where the maximum test duration was doubled, but this effect not due to the extended duration. Experiment 3 assessed reliability of a response to human approach at the farm rather than individual level, in this case assessing responses to an unfamiliar person. The proportion of calves making contact with the person was not reliable (rs = 0.22), but the proportion retreating from the person had moderate reliability (rs = 0.52). Reliability was improved by excluding data from calves that had coughs on the day of testing. Conducting multiple tests per individual using different stimuli and reporting health status of the animals are recommended for future research and animal welfare assessment schemes that include measures of fear.