Opinion of applied ethologists about expectation bias and debiasing techniques

Observers in applied ethological research are rarely blinded despite mounting evidence that expectation biases may affect, and even invalidate, research outcomes. We surveyed delegates of the ISAE-2014 congress shortly before (n=39 respondents) and after (n=51 respondents) a combined congress plenary lecture and workshop on expectation bias in applied ethology. During the lecture some examples of how expectancy effects influenced scores of animal behaviour and welfare were presented. The aims were to evaluate the effect of the plenary lecture and workshop on the opinion of the congress delegates on blinding observers and alternative debiasing techniques. Awareness about expectancy effects and debiasing techniques was lower before than immediately after the congress plenary lecture and workshop. Research situations considered as most susceptible to expectation bias – i.e. when the data-collector uses subjective methods and has strong expectations about the research outcome – were perceived to be more common in applied ethology than in other scientific disciplines (29.8 vs 16.9%, P<0.001). Nonblinded data collection in such research situations was viewed more disapprovingly after the plenary lecture and workshop as compared to before (lsmean of 6.8 vs 7.8 on a 10-point Likert scale, P<0.05). The main reasons why blinded observations are uncommon in applied ethology seem to relate to a limited awareness about expectancy effects and to logistic constraints of blinded observations rather than that the susceptibility of the research field is perceived to be low. In addition to the immediate effect of the plenary lecture and workshop, a more sustained and concerted effort throughout all stages of the research process seems warranted in order to avoid expectation bias affecting research findings and in order to safeguard the scientific credibility of the field of applied ethology