Online Program

Monte Carlo: An Underutilized Tool for Understanding Statistics

Keywords: Monte Carlo, simulation, applied probability

Monte Carlo simulation provides a simple, flexible, and intuitively appealing tool for understanding, and, perhaps more importantly, for explaining, statistical phenomena. Nonetheless, in our fondness for exact closed form solutions, or sophisticated numerical approximation, simulation techniques are often the Rodney Dangerfield of statistics. This talk proposes that Monte Carlo tools should see much greater use by practitioners. Moreover, we’ll argue that, in applications, simulated solutions are sometimes preferable to analytical ones by virtue of their speed and ease of use, and their communicability to statisticians and non-statisticians alike. We'll review a little of the history of Monte Carlo techniques that might explain its tepid acceptance by statisticians. Then we’ll use Monte Carlo techniques to provide explanations of some common phenomena in the manufacturing and engineering world that even many statisticians may not be aware of. The principle example addressed will be that of understanding differing customer and supplier perceptions of defect rates. Time permitting, we'll briefly discuss one or two other interesting uses, as well.