A gentle tutorial on power simulations in suicide research: Worked examples in R

Abstract Introduction Power analysis is critical for both planning future research samples and evaluating the reasonability of answers produced by pre‐existing and fixed samples. Unfortunately, the irregularity of suicide‐related data and the need for increasingly complex models in suicide research can make traditional power formulas inaccurate or even unusable. Ignoring these common problems risks both over‐ and under‐recruiting, as well as obscuring the true quality of the results (up and down) to future reviewers and readers. Method A better option is to use Monte Carlo power simulations. Results These techniques produce answers that are equivalent to traditional power formulas when traditional assumptions are met, but produce more accurate results in the common case when those assumptions are violated. Conclusion What follows is a tutorial on how suicide researchers can conduct such simulations. It begins by building the reader’s intuition for why simulations work, followed by two worked examples in R. Discussion also includes guidelines for conducting and reporting simulations, along with answers to frequently asked questions. Appendices provide code examples researchers can model and adapt to their own simulations as needed.