pharmacology-small-sample-EDA
Description
This repository contains the complete suite of R-scripts associated with the manuscript: "Beyond Bar Charts: A Practical Guide To Small Sample Pharmacological Data" (Sachs & Marino, 2026). An interactive version of the Confidence Funnel tool is hosted at https://mjmarino-phd.shinyapps.io/confidencefunnels
Files
Steps to reproduce
Overview In pharmacological research, experimental constraints often limit sample sizes to n=7 to 30. These scripts provide a practical, beginner-friendly toolkit for: *Outlier Detection: Applying Dixon’s Q-test. *Effect Size Estimation: Calculating Cohen’s d, Hedges’ g, and Cliff’s Delta. * Data Visualization: Generating high-quality, reproducible figures for publication. * In addition, the R-scripts employed to generate figures 1 - 9 of the manuscript are included with substantial annotation sufficient for the novice to use as an entry to R programming Repository Contents The repository includes 16 annotated R-scripts. If you are new to R, we recommend starting with: 1. Fig1_final.R: A simulation showing how small-sample stochasticity affects mean estimates 2. Fig2_final.R: A demonstration of the Central Limit Theorom in action and the basis for the n=30 'parametric pivot' 3. Fig3_final.R: A simulation of a log-normal population distribution and log transformation 4. Fig4_final.R: An overview of the anatomy of a Tukey Boxplot 5. Fig5_final.R: Application of Tukey boxplots and overlaid scatter plots to simulated experimental data 6. Fig6_final.R: Use of 'caterpillar' plots to demonstrate the concept of confidence intervals 7. Fig7_final.R: Application of caterpillar plots to simulated experimental populations 8. Fig8_final.R: Introduction of 'Confidence Funnel' plots as a novel data visualization tool 9. Fig9_final.R: A final summary figure suggested as an ideal method to report small data set results 10. GenerateSampleData.R: Script used to search for the type of data set used as a case study in the manuscript. 11. SimpleDixonQ.R: A highly annotated example of how the `dixon.test` function works in R 12. AutomatedDixonQ.R: A fully automated implementation of the outlier screen logic presented in the manuscript provided as a tool for analyzing data in CSV files formatted like `CaffeineData.csv` containing any number of groups 13. Cum_ProbabilityObservingOutlier.R: Application of the binomial theorem to estimate the probability of observing 1, 2, or 3 outliers based on sample size 14. EffectSIzeCalculator.R: Calculates Cohen's d, Hedges' g, and Cliff's Delta all with 95% CIs for data in CSV files formatted like `CaffeineData.csv` containing two groups 15. ConfFunnel_with_Points.R: Generates Confidence Funnel plots for data in CSV files formatted like `CaffeineData.csv` containing any number of groups 16. CaffeineData.csv: Data for case study in the manuscript. File is required for generating Fig 5,8-9 and serves as an example template for analysis scripts Requirements These scripts were developed using R (version 4.3.3 (2024-02-29)) and utilize the following libraries: `effsize` (for Hedges' $g$ and Cliff's Delta) `outliers` (for Dixon and Grubbs tests)
Institutions
- Drexel UniversityPennsylvania, Philadelphia