Speaker: Phil Chalmers, York University
Department of Psychology

Title: Introduction to Monte Carlo Simulations with Applications in R

Abstract: Monte Carlo simulation (MCS) methods are useful when studying the properties of statistical estimates and estimators. More specifically, MCSs are digitally-driven computer-intensive experiments used to generate and analyze plausible data-sets, and the results from these analyses are summarized using suitable meta-statistical methods. MCSs are invaluable to the field of statistics because they provide flexible techniques for studying the properties of statistical estimators under real-world conditions; for instance, how statistical estimates behave in small sample sizes, how robust models are when assumptions are violated, how to determine Power rates when no analytical formulae are available, and so on.

This talk will provide an introduction to MCS theory and methods, and will demonstrate how the R computing environment can be utilized when writing code for MCSs. However, a package specifically designed for MCSs (called SimDesign) will be discussed in greater depth, and the organization of this package will be contrasted with the more common 'for-loop' approach. Topics such as Bias, Efficiency, Coverage, Type I error rates, and Power will be explored, and various examples will be presented to help demonstrate how MCSs can be organized, summarized, and presented.