Can we measure human–animal interactions in on-farm animal welfare assessment? Some unresolved issues : Types of measures
An immediate unresolved issue is what type of measure to use. A number of different ways of measuring animals’ fear of people have been used. These can be roughly classified into those that measure the distance that animals keep between themselves and people (‘‘distance measures’’), those that assess the animals’ response to being handled (‘‘handling measures’’), or those that rely on some subjective rating of the animal.
The most common method involves taking some measure of how much the animals approach or avoid people (e.g. Hemsworth et al., 2000; Breuer et al., 2000; Waiblinger et al., 2002); the underlying assumption being that animals that are most fearful of people will keep the greatest distance. These measures are usually taken by placing the animal together with a person in a limited space and taking some measure of the extent that the animal avoids or approaches the person. A distinction is usually drawn between the extent that an animal will voluntarily approach a stationary person (which we will call ‘‘approach’’ distance), and the distance that an animal will allow a person to approach (which we will call the ‘‘flight’’ distance). Across different studies, there has been little effort to standardise how such measures are taken. For example, approach distance has
been measured either by the minimum or average distance that the animal approaches, the latency to make contact, or the amount of time the animal spends in contact or at various distances from the person.
A second type of test is to observe the animal’s responses when it is actually being handled. Responses to handling have been observed during normal or routine handling operations, such as milking for dairy cows (e.g. Hemsworth et al., 2000; Breuer et al., 2000), feeding of veal calves (Lensink et al., 2001), or during auction sales of beef cattle (e.g. Lanier et al., 2001). Since these tests include active handling of the animal, some standardisation of the handling is very important, otherwise variation in the handler’s skill or in the type of handling the animals receive could influence the results. However, an impressive diversity of tests have been used. With only a few exceptions, each research group appears to prefer to invent a new way of assessing reactions to handling rather than
adopting someone else’s techniques; thus, there is little standardisation of measures and it is difficult to compare and collate results across studies. For example, some researchers use a measure based on the time taken to actually complete the handling task (e.g. Lensink et al., 2000, 2001; Boissy and Bouissou, 1988; Boivin et al., 1992). Other researchers take
some measure that might indicate fear in the animals while they are being handled. These include behavioural measures, such as the incidence of kicking or stepping by dairy cows (Hemsworth et al., 2000; Breuer et al., 2000), running or kicking or escape attempts (Le Neindre et al., 1995; Lensink et al., 2001) or even acts of aggression against the handler
(Price and Wallach, 1990).
A third alternative is to obtain some subjective rating of the animal. Some rating scales are grounded on descriptions of observable behaviour (e.g. Voisinet et al., 1997; Lanier et al., 2001), while other rating scales describe the overall ‘‘personality’’ or ‘‘temperament’’ of the animal and the observers are asked to make a subjective rating of the animal, rather than recording what the animal is actually doing.
In principle, any one of these types of measures could be used in on-farm assessment, although ‘‘distance’’ measures have received most attention in on-farm welfare assessment (e.g. Main et al., 2003; Rousing andWaiblinger, 2004). However, the relationship between the different types of measures and the relative merit of each measure as an indicator of human–animal relationships has not been adequately examined. Nor has there been enough attempts to standardise how these measures are taken by different researchers. This limits our ability to use the accumulated research to upport the use of such measures in animal