Volunteer experience influences the conclusions of behavioural experiments

Volunteers offer an inexpensive and rapid means of collecting behavioural data, but their reliability is often overlooked. Past research has suggested that observers that were inexperienced are equally adept at recording behaviour as experienced observers, and inexperience was regarded as being merely unfamiliar with a sampling technique but not unknowledgeable about behaviour. The aims of our study were (i) to investigate the reliability of relatively naïve volunteers (i.e. those with no prior behavioural scoring experience) as behavioural data collectors; and (ii) to test the influence of the strength of inter-observer concordance on the outcome of testing a specific ethological hypothesis. Two cohorts of volunteers (high school and university students) conducted observations on a group of captive chimpanzees, simultaneously with an experienced observer (LD), recording behaviour and the location of the chimpanzees in their enclosure. Kendall's Tau agreement scores and odds ratios indicated poor agreement between inexperienced volunteers and the experienced observer, regardless of the educational experience of the volunteers and difficulty of the behaviour scored. We compared the data between the volunteers and experienced observer by independently testing each dataset with regard to the hypothesis that the chimpanzees were stressed by being in close proximity to the public. The school cohort data supported the hypothesis, while the time-matched experienced observer data suggested no relationship between public proximity and stress in chimpanzees. A separate analysis of the university cohort and time-matched experienced observer data both indicated that chimpanzees were more stressed at locations away from the public. These findings suggest that inter-observer agreement scores offer insights into the precision of data but not accuracy. Furthermore, the use of volunteers as data collectors should be assessed in relation to the aims of the study in question, since volunteers may be appropriate for studies of general patterns of behaviour but not for detailed ethological examinations.