Conference proceedings article
Validity aspects of behavioural measures to assess dairy cows´ responsiveness towards humans.

Publication Details
Ebinghaus, A.; Schmitz, L.; Ivemeyer, S.; Domas, L.; Knierim, U.
Newberry, Ruth C.; Braastad, Bjarne O.
Wageningen Academic Publishers
Publication year:
Pages range:
Book title:
Animal lives worth living - Proceedings of the 53rd Congress of the ISAE, 5th-9th August, 2019, Bergen, Norway
Title of series:
Animal lives worth living - Proceedings of the 53rd Congress of the ISAE, 5th-9th August, 2019, Bergen, Norway


assessment of the human-animal relationship (HAR) in cows, different measures
reflecting the animals´ behaviour towards humans are used. An established and widely
tested measure is the avoidance distance (AD). This test records the distance a
cow allows a person to approach. Further HAR measures applied relate to responses
in more intense human-animal interactions. For instance, cows´ tolerance to tactile
interaction (TTI), their behaviour during and after release from restraint (RB),
and body language during the human-animal interaction by means of qualitative
behaviour assessment (QBA) have been assessed in previous investigations, and
tested with regard to reliability and criterion validity. However,
uncertainties still exist regarding the measures´ construct validity,
vulnerability to confounders or risks of observer bias. These aspects are
addressed in the current study.

RB, and QBA were investigated on a research farm with 102 dairy cows regarding 1)
causal relationships with frequent positive human-animal interactions (6-days
period of concentrate provision by hand), 2) potential confounding effects of
lactation status (before/after calving), 3) potential short-term effects of management
procedures (claw trimming (CT)), and 4) the occurrence of expectation bias. Concerning
objectives 1-3, different samples of cows were assessed repeatedly and differences
in behaviour before and after hand feeding of concentrate, calving, and CT were
analysed by using paired Wilcoxon tests, for differences before and after CT
using blinded observation of video records. For objective 4, non-blinded live
and blinded video assessments before and after CT were carried out. Assessments
strongly correlated regarding all measures (analysed before CT; AD: Spearman
rank correlation (rs)=0.93, n=29; TTI: Cohen´s Kappa (К)=0.88, n=29;
RB: К=1.00, n=29; QBA: rs=0.96, n=15, all p<0.001), but QBA
scores ranged on different levels (paired Wilcoxon: p=0.027). Therefore, it was
tested separately for live and blinded video assessments whether assessments
before and after CT differed.

After the
period of hand feeding, the cows had lower ADs (p=0.013, n=27), responded less
fearful regarding TTI (p=0.045, n=27) and QBA (p=0.026, n=14), and by tendency
regarding RB (p=0.052, n=27), supporting the measures´ construct validity in
the way that they reflect an improved HAR after frequent positive interactions.
Neither (blinded) assessments before and one day after CT (p=0.489–1.0, n=29) nor
before and after calving (p=0.244–1.0, n=13) differed, suggesting that single routine
procedures and physiological changes around calving do not confound assessments.
Also non-blinded live assessments did not yield any significant differences
before and after CT (paired Wilcoxon: p=0.151-1.0, n=15-29). However, numerically
observers assessed cows as less fearful after CT in live QBA compared to
blinded QBA. Since observers were involved into the research question, they
might have tried to avoid bias. Although, we could not detect a significant
expectation bias, depending on the measures chosen, blinding should be considered
in studies involving certain expectations.

Last updated on 2019-15-08 at 15:39