Illegal use of natural resources is a threat to biodiversity globally, but research on illegal activities has methodological challenges. We examined 100 studies that empirically identify targeted resources, techniques used to procure resources illegally, locations of illegal activities, characteristics of typical violators, incentives driving illegal use of resources, magnitude of the problem of illegal use (e.g., quantities used), or frequency of illegal activity. We based our evaluation of the methods used in these studies on their ability to provide these empirical data, relative labor demands, training and technology requirements, and levels of uncontrollable bias. We evaluated eight different methods: law-enforcement records, indirect observation, self-reporting, direct observation, direct questioning, randomized response technique (a survey method designed to improve accuracy of responses to sensitive questions), forensics, and modeling. Different situations favored different methods, each with distinct advantages and limitations. Six context-specific factors-location of resource use (in situ vs. ex situ), budget, technology and training capacity, ease of detection of illegal activity, scope of illegal activity (limited vs. widespread), and researchers' willingness to accept bias in results-help narrow the choice of methods. Several methodological concerns applied to any study of illegal resource use: regular monitoring can detect trends; modeling can incorporate sampling error and data uncertainties; researchers must manage levels of bias that vary between methods; triangulation of results from multiple methods can improve accuracy. No method is a panacea, but a combination of techniques can help address the lack of data on illegal activity. Researchers empirically compared results from different methods in only four studies, and no one has compared more than two methods simultaneously. Conservation would benefit from more research focused on: methods comparisons that include cost effectiveness, time efficiency, and statistical rigor; unique applications of the eight techniques currently in use; and testing of new methods.
