Researching agricultural environmental behaviour: Improving the reliability of self-reporting

摘要

Agricultural practices cause many of the environmental problems in river basins. Changing farmer behaviour to adopt more sustainable practices is a key focus of government policy in many countries. There is now a need to assess the effectiveness of projects that promote environmental agricultural behaviour. Most agricultural research that evaluates landholder practices relies on farmers to report about their own behaviour. This behavioural measure, known as 'self-reporting', has been widely critiqued because reporting is often biased. Little is known about the reliability of self-reports about environmental behaviour, and even less is known about self-reporting agricultural environmental behaviour. This paper considers the extent that agricultural environmental research relies on self-reported data, presents a case-study comparing farmer self-reports with more reliable observed proxy data, and offers some methods for minimising self-reporting bias, particularly bias related to participant perceptions of social desirability. We compared self-reports about farmer environmental behaviour (preventing cattle from grazing riverbanks) with observed proxy data (e.g., visual evidence of cattle access) and found that more than 60% of self-reports were inaccurate, including both under- and over-reporting of grazing behaviour. We found that self-reporting is less reliable for identifying behavioural determinants compared to using observed proxy data. We also found that farmers experience social pressure to perform environmental behaviours. Thus, we suggest the inaccuracy of self-reported data may be the result of social desirability bias. Substantial investment has been made to assess the effectiveness of government policy for encouraging agricultural environmental behaviour. The success of such programs relies on the accuracy of behavioural data. Agricultural research often depends on self-reported data. Thus, researchers should make efforts to design projects to reduce the likelihood of self-reporting bias.