Characterizing noncompliance in conservation: a multidimensional randomized response technique for multinomial responses

Rule violation is critical for biological conservation worldwide. Conventional questionnaires are not suitable to survey these violations and specialized questioning techniques that preserve respondents privacy, like the forced-response RRT, have been increasingly adopted by conservationists. However, most of these approaches do not measure multinomial answers and conservationists need a specialized questioning technique for real-world settings where noncompliance could occur in different forms. We developed a multidimensional, statistically-efficient, RRT which is suitable for multinomial answers (mRRT) and which allows researchers to test for respondents noncompliance during completion. Then, we applied it to measure the frequency of the various forms of illegal restocking of European catfish from specialized anglers in Italy, developing an operational code for the statistical software R. A total of 75 questionnaires were administered at a large fishing fair in Northern Italy, in winter 2018. Our questionnaires were easily compiled and the multinomial model revealed that around 6% of respondents had moved catfish across public freshwater bodies and private ponds. Future studies should better address their characteristics, and the mRRT could allow for modeling the effect of covariates over restocking behavior. The multinomial mRRT could be adopted to measure many forms of rule violation in conservation that could take different forms, like various forms of fish restocking or different modes of wildlife persecution.