rk.bayesian

rk.bayesian is an external plug-in for RKWard that introduces a full suite of Bayesian statistical tests powered by the BayesFactor package. It is designed to provide "Jamovi-like" output, including Bayes Factors, posterior effect size estimates, descriptive statistics, and visualization plots.

Features

Evidence-Based Statistics

• Bayesian T-Tests: Calculate Bayes Factors ($BF_{10}$), Posterior Effect Size ($\delta$) with Credible Intervals, and Raincloud plots (Boxplot + Jitter + Density).
• Bayesian ANOVA: Calculate Bayes Factors for model comparison, group descriptives, and data plots.
• Bayesian Correlation: Test for associations ($BF_{10}$), estimate posterior Rho ($\rho$), and generate scatterplots with density curves.
• Bayesian Contingency: Test for independence in contingency tables using Independent (Poisson) or Joint Multinomial sampling.
• Automatic Interpretation: Results tables include evidence strength labels (Anecdotal, Moderate, Strong, etc.) based on Jeffreys' scale.

🌍 Internationalization

The interface is fully localized to match your RKWard language settings:

• 🇺🇸 English (Default)
• 🇪🇸 Spanish (es)
• 🇫🇷 French (fr)
• 🇩🇪 German (de)
• 🇧🇷 Portuguese (Brazil) (pt_BR)

Dependencies

This plugin relies on the following R packages:

• BayesFactor (Bayesian inference)
• dplyr (Data manipulation)
• ggpubr (Plots)
• rkwarddev (Development)

Installation

Installing from GitHub

You can install the latest version directly from GitHub using the devtools or remotes package in R:

1. Execute the commands printed in the console:

library(devtools) # or library(remotes)
# install_github("AlfCano/rk.bayesian")
install.packages("rk.bayesian", repos = c('https://alfcano.r-universe.dev', 'https://cloud.r-project.org'))

2. Restart RKWard.

Usage Guide & Tutorials

The plugin adds a new entry to the main menu under: Analysis > Psychology / Social Science > Bayesian

Phase 1: Data Preparation

To follow these tutorials, please run this code in the RKWard R Console to load the necessary datasets:

# Load standard datasets
data(ToothGrowth)  # For T-Tests
data(PlantGrowth)  # For ANOVA
data(mtcars)       # For Correlation

# Create a Matrix for Contingency Table test
my_matrix <- matrix(c(20, 5, 10, 30), nrow = 2, 
                    dimnames = list(c("Male", "Female"), c("Yes", "No")))

Phase 2: Bayesian Independent T-Test

Goal: Check the strength of evidence for the difference in tooth length (len) by supplement (supp).

1. Menu: Psychology / Social Science > Bayesian > Bayesian Independent T-Test

Screenshot of the path in the main menu.

2. Dependent Variable: len.

Screenshot of the RKWard interface running the Bayesian T-Test module and setting the dependent and independent variables.

3. Grouping Variable: supp.

4. Options: Check "Posterior Effect Size" and "Plot Posterior Density".

Screenshot of the RKWard interface running the Bayesian T-Test module and setting the dependent and independent variables with preview.

5. Submit.

• Result: A table showing the Bayes Factor ($BF_{10}$), the Median Posterior Effect Size (Cohen's d) with 95% Credible Intervals, and a plot of the posterior distribution.

Screenshot of the output in HTML format, showing the results of the Bayes Factor ($BF_{10}$), the descriptive table and the corresponding graphic rendering.

Phase 3: Bayesian One-Way ANOVA

Goal: Compare models for plant weight (weight) across groups (group).

1. Menu: Psychology / Social Science > Bayesian > Bayesian One-Way ANOVA
2. Variables: Select weight (DV) and group (Factor).
3. Options: Check "Descriptives" and "Plot Data".
4. Submit.
   • Result: A table comparing the group model against the Intercept-only model with evidence labels, plus a table of group means/SDs.

Phase 4: Bayesian Correlation

Goal: Test association between MPG and Horsepower.

1. Menu: Psychology / Social Science > Bayesian > Bayesian Correlation
2. Variable 1: mpg.
3. Variable 2: hp.
4. Options: Check "Plot Scatterplot".
5. Submit.
   • Result: The Bayes Factor for the correlation, posterior estimate for Rho, and a scatterplot with regression line.

Phase 5: Bayesian Contingency Table

Goal: Test independence of rows/columns in a matrix.

1. Menu: Psychology / Social Science > Bayesian > Bayesian Contingency Table
2. Contingency Table: Select my_matrix.
3. Sampling Type: Independent (Poisson) or Joint Multinomial.
4. Submit.
   • Result: A Bayes Factor testing the independence of the rows and columns.

Author

Alfonso Cano
Benemérita Universidad Autónoma de Puebla
License: GPL (>= 3)

---

This plugin was developed with the assistance of Gemini, a large language model by Google.