Abstract:
|
In December of 2010, Justice Eileen Bransten ruled that statistical sampling was an acceptable form of legal evidence. This capstone ruling cemented the emerging trust and value of statistical estimation in pursuing the public good within a courtroom. With authority comes responsibility – a concept which implies the existence of standards. Using an evaluation framework of 5-transcendent criteria, we launched a robust system of standards that paved the way for a 3-fold increase in our productive output. Null-rich sample sets such as audit samples have long been found to distort or defer the effects of the central limit theorem. To overcome this, we profile the empirical performance of sample designs through simulation and this becomes are an important part of the testing framework. The insights provided by controlled simulation under targeted measures has proved not only to control reliability, but to provide a powerful catalyst for creativity and innovation. We will present the criteria, software routines and examples of how this supports innovation in sample design.
|