|
Abstract:
|
Simulation techniques are beneficial in statistical planning and computation. The R programming language provides numerous methods for generating random variables. Nonetheless, simulation studies can grow complex in terms of the coding, dependencies in variables, repeated experiments, and associated analyses. To mitigate these concerns, the authors developed an R package called simitation.
The proposed methods are intended to simplify the development and investigation of simulation studies. Settings such as randomized controlled trials can be simulated and analyzed from a single line of code. For multivariate studies, users can generate data by specifying symbolic inputs for common distributions and regression-based dependencies. Empirical analyses of simulations, such as the distribution of estimated effects, false positive rates, or the statistical power, can be implemented across repeated experiments for customized settings. The presentation will focus on use cases for the software and present examples of how simulations can be easily developed.
|
